Compare commits

...
256 Commits
Author SHA1 Message Date
sam 474fd59759 chore(release): v1.54.0
deploy / gate (push) Canceled after 0s
deploy / archied-image (push) Canceled after 0s
deploy / agent-image (push) Canceled after 0s
deploy / dist-zip (push) Canceled after 0s
2026-10-16 03:00:32 +11:00
sam 40ee9ffafa fix(rpc): bound gRPC servers and return listener failures
Cap concurrent streams, the handshake and dead peers' streams on every
service, drain shutdown for at most 10s, and stop messaging when its
listener dies instead of running on without Deliver.
2026-10-16 02:57:26 +11:00
sam 939b912889 refactor(ui): unify account administration under Access 2026-10-16 02:57:13 +11:00
sam 83aa17cf8b feat(identity): add browser setup and user accounts 2026-10-16 02:41:24 +11:00
sam 9cdae5642e feat(rpc): mTLS between services from operator-issued certificates
Replace the service bearer tokens with [tls] ca_file and cert_dir. Caller
identity comes from the verified certificate, not metadata. Without TLS,
services listen and dial on loopback only. Task containers get only the CA
and keep their grant.
2026-10-16 02:37:51 +11:00
sam dcf2972f26 feat(ui): start tasks from the dashboard and slim it to activity and setup 2026-10-14 23:28:37 +11:00
sam f96c44d092 refactor(scheduling): remove unused engine accessors 2026-10-14 23:10:35 +11:00
sam f537d5f8cd refactor: delete wrappers superseded by live entry points
Remove controlplanerpc.NewClient (superseded by NewRPCClient), Loader.Overlay
(-config-overlay reaches Resolve directly), and setup.Run (RunParams is
the live entry point). Move ResolverFunc into a test file, since only
tests adapt functions to the egress Resolver.
2026-10-14 23:10:35 +11:00
sam c833b651a9 fix(workflow): set BuildNoChanges from writable stages and clear it on commit
A writable agent stage that passes now marks BuildNoChanges; repo.commit
clears it when the stage commits changes. The flag had no setter, so the
signal was lost in the YAML migration. The outcome of an empty commit
stays with if_empty: unset still fails, complete closes the issue, skip
moves on.
2026-10-14 23:10:34 +11:00
sam 062f5332a0 refactor(gateway): delete unused wrappers and in-memory session store
Remove gateway re-exports of messaging helpers with no callers, the
unused DeltaFunc adapter, DefaultPersonas, NewStoreTaskCreator, the
SortBySeq stub, and the in-memory session store that nothing constructs.
2026-10-14 23:10:34 +11:00
sam d9fe384fd9 build(deps): bump ai-sdk v0.1.43 -> v0.3.0 and update dependencies
Gemini 3.x/4 chat now routes through the Interactions API transport, which
replays thought signatures on tool turns; long decode stays opt-in. Also
takes go get -u updates (x/*, genproto, goose, nats-server, modernc sqlite).
2026-10-14 23:10:34 +11:00
sam 8a7220f5e9 fix(gateway): isolate model selections by org 2026-10-14 21:19:34 +11:00
sam 5b9ff6fa47 fix(gateway): restore provider model selection in chat
Restore provider groups and concrete model references in chat snapshots. Permit catalog selections while retaining named aliases and tenant model restrictions. Show the active model in /profile without requiring a task identity.

Refs: archie-core-mvek6
2026-10-14 21:03:30 +11:00
sam 66b22f4833 build: trim filesystem paths from service and agent binaries 2026-10-14 03:49:31 +11:00
sam acec5d392d feat(archied): read-only run-scoped sandbox view over the harness contract (#1662)
* feat(archied): read-only run-scoped sandbox view over the harness contract

Serve RunTarget: the daemon resolves a running task and attempt to its Kit
container and follows its output, refusing a finished or unknown task with a
stated reason. The dashboard adds a Sandbox tab behind the task log's read
authority. Takeover stays out of scope until the sod1 authority is wired.

* fix(archied): scope and close sandbox views correctly

Forward harness activity into container output, bind containers and dashboard views to their attempt, enforce the asserted org, and report unavailable dependencies and interrupted streams accurately.

Refs: archie-core-qi91, archie-core-252tj
2026-10-14 02:57:32 +11:00
sam 59be81c7f6 fix(webui): an absent setting equals its zero value in drafts; setting rows get roomier padding 2026-10-13 20:20:03 +11:00
sam c0ac39b8da fix(webui): the danger setting row tints without shifting its content 2026-10-13 19:55:05 +11:00
sam 7b19935a14 build(webui): add the CardAccordion chunk missing from the bundle 2026-10-13 19:49:25 +11:00
sam fa7d0d83c1 fix(archied): report schedules apply status at start and on a timer 2026-10-13 19:08:55 +11:00
sam 7542e3b15e fix(archied): report workflow definitions apply status at start and on a timer 2026-10-13 19:06:53 +11:00
sam 6412e6d116 feat(webui): settings lists share a card accordion that opens the new item and blocks adding while unsaved 2026-10-13 19:06:01 +11:00
sam b1dbe519b0 feat(webui): repositories list is an accordion; adding opens the new one and waits for save 2026-10-13 19:04:14 +11:00
sam 3103f87b71 fix(applystatus): the gateway no longer owes an apply status for channel and plugin settings 2026-10-13 19:00:52 +11:00
sam 8664e8cd97 fix(webui): drop settings hints that restate their label 2026-10-13 18:11:12 +11:00
sam 9e566e7278 fix(webui): shorten route descriptions, setup steps, alerts and panel errors 2026-10-13 18:10:39 +11:00
sam f02a3b6d02 fix(agent): drop the one-question cap from the chat prompt and SOUL 2026-10-13 17:49:35 +11:00
sam bf6f3512b1 feat(agent): default persona and SOUL share a reporting-focused identity prompt 2026-10-13 17:47:46 +11:00
sam 00112ae205 fix(webui): attention vocabulary no longer addresses a single user 2026-10-13 17:42:11 +11:00
sam 8001c893bd fix(webui): cut tutorial and marketing prose from labels, hints, empty states and confirms 2026-10-13 17:37:53 +11:00
sam 0e25291d47 fix(webui): review settings drop the in-force echo and rewrite the approval hint 2026-10-13 17:28:31 +11:00
sam 68fb3a7e19 fix(webui): form fields sit on the card colour so they no longer read as disabled 2026-10-13 17:24:48 +11:00
sam 541575a073 fix(archied): load and report EDA playbooks at start instead of at the first dispatch 2026-10-13 17:22:01 +11:00
sam d4d24cb1db fix(webui): default-alias warning states the action only 2026-10-13 17:17:06 +11:00
sam 20cb2fd99d fix(webui): task page drops filler copy, the unattributed-events note and the New run tooltip 2026-10-13 17:16:25 +11:00
sam 4fd6efb7f0 fix(webui): alerts drop UI-explaining prose and align Dismiss with the title 2026-10-13 17:14:43 +11:00
sam 9eabc624ee fix(config): a task snapshot carries only the model limits its aliases resolve to 2026-10-13 17:07:40 +11:00
sam 01418b7e81 fix(webui): harness terminal-unavailable message says what the operator can do 2026-10-13 17:02:45 +11:00
sam cac763a339 feat(webui): finish redesign dashboard task and secret settings 2026-10-13 16:41:35 +11:00
sam 0dde93e57e fix(taskactions): a failed callee cancel no longer leaves the caller's run going
Issue: archie-core-kig5j
2026-10-13 16:22:40 +11:00
sam 7435b6efe5 fix(eda): the undispatched-capture batch lists only captures an enabled binding still owes
Issue: archie-core-kfjz1
2026-10-13 16:21:21 +11:00
sam 796c7a711b fix(postgres): age retention spares captures still waiting to be dispatched
Issue: archie-core-jrh2m
2026-10-13 16:19:12 +11:00
sam 4ac9fd852f fix(postgres): assign bootstrapped agent identities to the default org on every start
Issue: archie-core-n5yj5
2026-10-13 16:18:19 +11:00
sam 06d6293a9d fix(controlplane): instance settings are written by the system org only
Issue: archie-core-f5zso
2026-10-13 16:17:38 +11:00
sam 0c904fb9ef fix(controlplane): playbook dispatch resolves and caches per the task's org
Issue: archie-core-5t4z7
2026-10-13 16:16:31 +11:00
sam 59645969ee fix(e2e): repository policy applies live, so seeding no longer restarts archied
Issue: archie-core-m5aqb
2026-10-13 16:15:07 +11:00
sam acacdfdab5 fix(statestore): an invalid stored policy raises an alert and resolves when it validates
Issue: archie-core-kgxw.16
2026-10-13 16:14:16 +11:00
sam 2090c41960 fix(ops): external NATS needs a token and publishes monitoring on loopback only
Issue: archie-core-ov11
2026-10-13 16:09:23 +11:00
sam e5a67b8ff0 fix(webui): a source create is authorized against the workspace it names
Issue: archie-core-ed8tw
2026-10-13 16:00:37 +11:00
sam 8e6dfd785f fix(webui): channel reload route is authorized as the channel action
Issue: archie-core-4laa1
2026-10-13 15:59:36 +11:00
sam b9547b01a8 fix(staterpc): an org-scoped caller cannot read or write another org's harness secret
Issue: archie-core-jo15w
2026-10-13 15:58:51 +11:00
sam 0720e9629c fix(statestore): a restart never resolves a service alert still inside its grace
Issue: archie-core-07lyl
2026-10-13 15:56:23 +11:00
sam bb67bd77f1 fix(statestore): service-unhealthy grace counts from the last stored report and covers silent services
Issue: archie-core-07lyl
2026-10-13 15:29:29 +11:00
sam 8ecda341e0 fix(servicekit): report apply status only for kinds the process re-applies live
Issue: archie-core-mt91
2026-10-13 15:28:43 +11:00
sam dee78a876a fix(postgres): clamp event, token and denial limits to int32
Issue: archie-core-g14d
2026-10-13 15:27:36 +11:00
sam 33ccfb2505 fix(daemon): run-time workflow check sees the binding event's signature and address
Issue: archie-core-bmro8
2026-10-13 15:10:31 +11:00
sam 0330a65054 fix(postgres): confine identity writes and GetPrincipal to the caller's org
Issue: archie-core-wkey2
2026-10-13 15:07:48 +11:00
sam 4eb9c20c0c fix(workflows): callees of a test run inherit the test source 2026-10-13 12:46:48 +11:00
sam 1f4910a651 feat(workflows): test-run unsaved drafts and browse workflow version history
A test run pins a validated draft, runs with forge and push writes
suppressed and recorded per step, and is excluded from workflow stats.
The History tab lists revisions that changed a workflow, diffs them as
YAML and on the canvas, restores into the editor, and ties runs to the
revision they used.
2026-10-13 12:27:40 +11:00
sam 150bfd4605 fix(postgres): quote the alert lock key so a NUL cannot reach Postgres text 2026-10-13 03:15:06 +11:00
sam 3bf64933b0 fix(ui/topbar): replace Jump to with a command palette, drop filler chrome (#1650)
* fix(ui/topbar): replace Jump to with a command palette, drop filler chrome

Jump to opens a shadcn-vue Command dialog that finds pages, tasks and
workflows by name. The connection indicator shows only when the stream is
lost, the dead documentation link is gone, and the profile menu label has
no stray whitespace.

* fix(ui): trim stray whitespace from template text and drop em dashes from copy

Element text that started or ended with a line break rendered with leading
and trailing spaces. Also removes the unused 'soon' nav state and the nav
group components no route reaches.

* fix(ui): drop the space a hidden spinner left before button text

* fix(ui/status): one dense Status page, full-width settings, problems-only apply status

Status drops its tabs and the duplicate health card; services list only
what is wrong. Settings pages use the full width, the resizable sidebar
handle is gone, and apply status shows only processes that have not
applied a change. PageHeader puts the title and actions on one line.

* refactor(ui): use shadcn ToggleGroup and Badge, drop SegmentedControl and StatusPill

Status badges show only when a state needs attention: no Running, Enabled,
Active, Listening or live pills next to healthy rows. Removes the unused
pagination, radio-group and scroll-area components.

* fix(ui/history): use the shadcn Table with fixed columns so nothing is clipped

* fix(ui/workflows): apply ?workflow= only once the workflow is listed

Selecting it before the definitions loaded opened an empty canvas.

* refactor(ui/workflows): use shadcn Input, NativeSelect, Checkbox and Button in editors

Replaces hand-rolled inputs, selects, the repository mode switch and icon
buttons in the workflow editor, bind dialog, harness page, sources panel
and review gate. Adds the upstream shadcn-vue NativeSelect.

* style(ui): rounded-rectangle controls instead of pills

Halves the radius scale so inputs, selects and buttons stop reading as
pills, and badges and chips lose their full rounding.

* fix(ui/alerts): centre the bell's failed state and stop it contradicting Needs you

* fix(ui/tasks): one header line, aligned columns, no placeholder issue numbers

Tasks puts its filters beside the title and drops the stat tiles the
dashboard already shows. Sort headers align with their cells, chat tasks
no longer show their synthetic issue number, and Logs and the task
not-found page use the shared header.

* fix(ui/dashboard): shared header and cards, empty cards stay silent

Needs you uses the shadcn Card and drops its duplicate Open button; quality
gates and token outlook render only when there is something to show; the
throughput strip moves into the dashboard, its only user.

* refactor(ui/stores): one Pinia store per feature

Splits the site-wide control-plane store and the module-level refs into
settings, workflows, tasks, events, org and session stores. Identity
requests move into lib/api. Drops store fields nothing read: per-resource
stream state, the states map and loadApplyStatus.

* fix(ui/org): one header for the org area, sections drop their repeated titles

The org name is the page title with the admin's org switcher and New org
beside it; Members and Workspaces tables sit in cards like other pages.

* fix(ui/events): shadcn Tabs for the capture, map and bind steps

Replaces the numbered step strip with line tabs that carry each step's
count; the Listening pill is gone and Capture off shows only when it is.

* fix(ui/workflows): confirm a workflow delete in an AlertDialog, not window.confirm

* style(ui/settings): denser setting rows
2026-10-12 04:40:21 +11:00
sam 94d2f98cc0 feat(e2e): local stack, seed and browser checks for release validation 2026-10-12 04:28:26 +11:00
sam 2df2f72fc2 fix(ci): guard apparmor userns sysctl in deploy workflow 2026-10-12 04:17:11 +11:00
sam 8a3ba32c5a Feat/operator alerts (#1635)
* feat(state): durable operator alerts owned by the State Store

Alerts dedup on org+kind+subject, resolve themselves when a condition
clears, and are raised, resolved, dismissed and listed through the state
contract. Task transitions raise needs-human and failed alerts in the same
transaction; the store checks service health and prunes closed alerts.
Each raise and resolve queues a webhook delivery whose attempts are kept.

* feat(notify): deliver operator alerts to the webhook; raise config and release alerts

notify.webhook now receives raised and resolved alerts above a severity
threshold with HMAC-signed payloads, bounded retry and every attempt recorded
on the alert. archied raises model alias, provider and credential-binding
alerts from the effective configuration, an update alert from the release
channel, and a credential-rejected alert from the egress resolver. The
dashboard API lists and dismisses the caller's org's alerts.

* feat(ui): alert bell in the top bar and webhook alert settings

The bell lists the org's open alerts by severity with dismiss and a link per
alert, and refreshes from the event stream. The task settings page gains the
webhook severity threshold and signing secret. ui/dist rebuilt.
2026-10-12 03:30:46 +11:00
sam f34f60f9c7 Feat/org epic (#1634)
* fix(postgres): scope OpenPRs to the caller's org

A principal-carrying caller read every org's pr_open tasks. The query now
honours the request scope; the daemon's unscoped review scan still sees all
orgs.

Issue: archie-core-kgxw.23

* fix(statestore): serve task logs only to the task's org

ReadTaskLog and StreamTaskLogContent took a task ID and read the file with
no org check. The State Store's task-log reader now confirms the task is
visible under the caller's scope first; unscoped services read as before.

Issue: archie-core-kgxw.24

* fix(postgres): scope ListPolicies to the caller's org

Every org's policy text reached any principal-carrying caller. A scoped read
now sees its own org's policies, plus instance policies when scoped to the
system org. The engines and the lockout guard decide for every org, so they
read the full chain unscoped: the dashboard engine loads without the
request's principal and the guard lifts the caller's scope.

Issue: archie-core-kgxw.25

* fix(postgres): scope identity reads to the caller's org

ListIdentities and GetIdentity (and its alias fallback) returned every
identity to any caller. A caller scoped to an org now sees only that org's
members and assigned agents; the system org and unscoped services still see
the whole instance.

Issue: archie-core-kgxw.26

* fix(postgres): hide instance status from other orgs

ListPresence, ListChannelStatus, ListApplyStatus and GetConfigSnapshot
answered any org principal with instance-wide state. They now return nothing
unless the request is unscoped or scoped to the system org. The dashboard
reads the published config as itself, since its handlers project it for
callers of every org.

Issue: archie-core-kgxw.27

* feat(postgres): assign new agents an org and role at creation

A bot or service account created through the API served no org until the
next State Store boot granted it one. Create now assigns it to the org the
request acts in with the developer role in the same transaction, so it is
authorized immediately.

Issue: archie-core-kgxw.14

* feat(edastore): create event records in a chosen workspace

Sources are created in a workspace of the caller's org, the default one when
none is named; event types and bindings take their source's workspace and
mappings their event type's, the same rule the org backfill applied. The
binding, mapping and event type contracts carry the workspace, the four list
endpoints accept ?workspace=, and the Events page has a workspace selector
that filters every step and picks where a new source goes.

Issue: archie-core-kgxw.11

* feat(webui): authorize a record in its workspace

The API built every access request with no workspace, so workspace
policies and per-workspace roles never applied. A request naming a source,
binding, mapping, event type or task now reads the record's workspace under
the caller's scope and authorizes in it; tasks carry their workspace on the
State Store contract for this.

Issue: archie-core-kgxw.9

* feat(access): expose workspace environment to policies

Policies can test resource.workspace.environment. The workspace attribute
is the workspace entity, which carries its environment; the principal
brings its org's workspaces from the State Store, so the dashboard and the
daemon's run checks describe a resource's workspace without another lookup,
and the daemon now authorizes runs in the task's or binding's workspace.

Issue: archie-core-kgxw.10

* fix(webui): ask read_secret when a source secret is read

Rotating a source's secret returns its value, but the route asked update,
which every developer holds, so no policy could narrow who reads a secret.
It now asks read_secret on the secret, which the shipped roles give admins
and an object policy can narrow further.

Issue: archie-core-kgxw.17

* refactor(org): delete the unused upgrade-incomplete sentinel

Nothing returns org.ErrUpgradeIncomplete: the State Store finishes the org
upgrade before it serves.

Issue: archie-core-kgxw.18
2026-10-12 03:24:04 +11:00
sam 8540c9f7c8 feat(releaseupdate): transactional self-update through archied (#1633)
archied update install verifies the release zip, checks every binary
reports the release, pulls the agent image by digest, backs up the
database in-process and stages the binaries before handing off to
archied update watchdog in its own process. The watchdog restarts units
in release order, requires archied healthy for 120s and otherwise rolls
back binaries and image policy, restoring the database only when a
migration ran. Boot recovery resumes the frozen watchdog.

Deletes scripts/archie-update-install, scripts/archie-update-watchdog,
the updater's --from-source and --auto, CommandInstaller,
ScriptInstaller and the update-image and update-topology subcommands.
2026-10-12 03:21:02 +11:00
sam 3014db85e7 Fix/workflow eda audit (#1627)
* fix(webui): accept the merge task action

The action whitelist omitted merge, so the dashboard's Merge button on a
pr_open task was refused with 400 unknown action.

Closes: archie-core-slpoz

* fix(workflow): give each parallel branch its own outputs

Branches shared the task's Outputs map, so concurrent declared-output
writes raced (or were lost when the map started nil). Each branch now
writes a copy; join merges new outputs in branch-name order and refuses
one written by two branches as a second write.

Closes: archie-core-3a7w2

* fix(postgres): dispatch captures oldest first and keep pending ones

ListUndispatchedCaptures ordered newest first and listed captures no
binding can match (failed HMAC), so sustained intake starved older work
and bad signatures held the batch. It is now oldest first and lists only
dispatchable captures. Count retention no longer evicts a capture that
is still pending for an armed binding. The daemon lists only sources
with an enabled workflow, so captures waiting on a disabled one do not
hold the oldest-first batch.

Closes: archie-core-2et4b, archie-core-o0s0z

* fix(ui): refuse renaming a workflow onto an existing id

Saving upserts by id, so editing a workflow's id to another workflow's
id removed the open one and overwrote the other. The editor now refuses
that save and says why.

Closes: archie-core-wpoz8

* fix(workflow): reject unknown step settings at save

Step factories decoded settings with yaml.Node.Decode, which drops keys
it does not know, and settings-less steps ignored settings entirely, so a
misspelt key (readOnly for read_only) saved and silently did nothing.
Every factory now decodes strictly and settings-less steps refuse
settings; ParseDefinition runs the factories, so the save refuses them.

Closes: archie-core-8b5nk

* fix(postgres): keep a stored capture when its prune fails

InsertCapture committed the capture, then returned the prune's error, so
intake answered 500 and a sender without a delivery header retried into
a duplicate capture and a duplicate dispatch. A failed prune is now
logged and the stored capture's id returned.

Closes: archie-core-3miuh

* fix(crondelivery): fail and advance a job no runner can deliver

Router.Run returned nil for a kind with no runner, so the engine reported
a successful run while the schedule never advanced and the job fired on
every tick. The run now fails with the refusal, which the engine reports,
and the schedule advances (a one-shot is removed). The router's own
refusal event is gone with it.

Closes: archie-core-s6ahh

* fix(taskactions): stop the runs of cancelled callees

CancelExecution cancelled a caller's waited-on callees in the store, but
only the caller's run was stopped, so callee containers ran to the end.
CancelExecution now returns the callees it cancelled (nested included)
in place of the step ids nobody read, and every cancelling action stops
their runs too.

Closes: archie-core-9labz

* fix(workflow): treat a zero json.Number as false in when

Task inputs and step results decode numbers as json.Number, which truthy
did not unpack, so when: inputs.count held for a count of 0.

Closes: archie-core-uv270

* fix(webui): read a workflow's runs with a workflow-filtered page

The workflows page and the canvas run picker filtered the newest 100
tasks of every workflow client-side, so on a busy instance a workflow's
older runs were invisible. GET /api/tasks takes a workflow filter, carried
through the State Store's Tasks request to the query, and both views ask
for the selected workflow's runs.

Closes: archie-core-tpiw1

* fix(ui): load the restored workflow into the editor

Restore Defaults saved the shipped definition but left the open draft in
the editor, still marked unsaved, so a later save re-applied it. The
editor now reloads the restored definition.

Closes: archie-core-85e7w
2026-10-12 03:20:00 +11:00
sam aea7c5c9f1 build(tools): provision PostgreSQL 18 in the cloud setup for full checks 2026-10-12 03:00:46 +11:00
sam c792c552cf build(tools): link cloud setup tools onto the default PATH 2026-10-12 02:44:11 +11:00
sam a71f1e0817 ci: drop the Postgres client from the deploy gate 2026-10-12 02:19:51 +11:00
sam 76905c8ec6 ci: run every job on the self-hosted runner 2026-10-12 02:10:14 +11:00
sam 85d8b771f0 Merge pull request #1602 from samcharles93/feat/workflow-runs-as
Feat/workflow runs as
2026-10-12 02:08:30 +11:00
sam d31a950daa feat(workflow): shipped workflows run as the instance root 2026-10-12 01:53:55 +11:00
sam f765e23812 feat(daemon): run each workflow as its declared identity
Every attempt checks that the identity that started the task and the
identity the workflow declares may both run it, recording a denial for
either. A workflow that declares no identity is refused rather than
running as its starter or the root. A workflow.call callee is started by
its caller's declared identity instead of inheriting the caller's row.
The workflow settings panel marks the identity as required and offers
the agent identities.
2026-10-11 14:49:38 +00:00
sam d66dca44e0 Merge pull request #1601 from samcharles93/fix/atomic-dispatch
Fix/atomic dispatch
2026-10-12 01:29:16 +11:00
sam b8f8f9a272 fix(postgres): allocate synthetic issue numbers from a per-repo counter
InsertChatTask and EnqueueCallTask computed MAX(issue_number)+1, so
concurrent inserts in one repository took the same number and the second
failed on the unique key. Both now bump a chat_issue_numbers row, seeded
from the existing maxima, whose row lock serializes allocation.
2026-10-11 14:21:27 +00:00
sam e128f82e11 build(ui): install dashboard deps from the lockfile without rewriting it 2026-10-12 01:21:16 +11:00
sam daab5732e8 fix(daemon): claim binding dispatch and enqueue its task atomically
EnqueueBindingTask now takes the capture and claims the ledger row,
inserts the task, stamps its provenance and links the ledger row in one
State Store transaction. A failure anywhere leaves the delivery
undispatched for the next cycle; a replay or concurrent claim is refused
by the ledger's unique index. SetDispatchTask is removed.
2026-10-11 14:19:43 +00:00
sam eb5703b7dd ci: drop the Postgres client the removed tests needed 2026-10-12 01:17:11 +11:00
sam ce5fc3e226 test(postgres): remove database-backed tests and the testcontainers harness 2026-10-12 01:17:02 +11:00
sam c2a649ded4 docs(agents): forbid database-backed tests 2026-10-12 01:15:19 +11:00
sam 1b0a89ca2e fix(tools): keep bd hook setup out of the checkout 2026-10-12 01:14:38 +11:00
sam 1441585e06 chore: reindent beads config and add agent profile 2026-10-12 01:01:19 +11:00
sam f3576342b6 chore(tools): rely on preinstalled gh in cloud setup 2026-10-12 00:46:51 +11:00
sam ede2102c60 chore(tools): install gh in cloud setup 2026-10-12 00:43:37 +11:00
sam 16d98dd47c chore(tools): install bd via upstream installer and disable attribution in cloud setup 2026-10-12 00:41:07 +11:00
sam 7aca31b18f chore(tools): add Claude cloud environment setup script 2026-10-12 00:12:18 +11:00
sam e97f1fa609 fix(ci): limit gate concurrency to runner memory 2026-10-11 22:28:11 +11:00
sam 62f0baba36 fix(ci): use repository-scoped quality runner 2026-10-11 22:10:52 +11:00
sam 40fa6f9d7d fix(ci): check namespace support before installing tools 2026-10-11 21:33:56 +11:00
sam 6ce605d6ea fix(ci): allow extension namespaces in quality container 2026-10-11 21:30:43 +11:00
sam feba855326 style(ui): tidy org page layouts and empty states
Members and Workspaces get fixed column widths and their add forms move
into labelled sections. Agents lists agents first and moves credential
grants below. Access puts New policy in the card header. Tokens states
when none exist.
2026-10-11 15:59:30 +11:00
sam 47f75f71a7 docs: remove incorrect notice 2026-10-11 12:17:31 +11:00
sam 4e060ce5ef fix(ui): stage history restore as a draft applied from the save bar 2026-10-11 11:35:02 +11:00
sam e2e24f9d56 ci(deploy): pull the container image the uncached test suite needs
deploy / gate (push) Canceled after 0s
deploy / archied-image (push) Canceled after 0s
deploy / agent-image (push) Canceled after 0s
deploy / dist-zip (push) Canceled after 0s
2026-10-11 04:26:08 +11:00
sam 5aa5c2a6c8 chore(release): v1.53.0 2026-10-11 04:23:10 +11:00
sam 41b2daed0c Merge branch 'work/53jc' 2026-10-11 04:11:43 +11:00
sam e0a79b3990 Merge branch 'work/ry4k' 2026-10-11 04:11:43 +11:00
sam e2d6b75b10 fix(prreview): restore the benchmark runner on configured models and public PRs 2026-10-11 04:11:40 +11:00
sam b4be999777 fix(gateway): remove the web /restart command that no process could serve 2026-10-11 04:11:37 +11:00
sam 5c3bf0b642 feat(storepkg): auto-update policy, applied by a State Store ticker 2026-10-11 04:10:53 +11:00
sam be2b897149 refactor(servicediscovery): delete unreachable contract and NATS implementation 2026-10-11 04:09:56 +11:00
sam f5a39cfb9b fix(daemon): bound a package workflow's triggers by its accepted authority 2026-10-11 04:08:02 +11:00
sam f5a6fa5b32 fix(postgres): number the outcome-posted migration after package updates 2026-10-11 04:04:45 +11:00
sam 93ec6d916b Merge branch 'work/mt9xo' 2026-10-11 04:03:52 +11:00
sam c36fb29683 Merge branch 'work/zcg8' 2026-10-11 04:03:52 +11:00
sam cac0f43b58 Merge branch 'work/mqti13'
# Conflicts:
#	internal/contracts/state/v1/state.pb.go
2026-10-11 04:03:52 +11:00
sam 805364f201 feat(archied): post a chat task's outcome back to its chat 2026-10-11 04:03:30 +11:00
sam 3b2a3a7630 fix(webui): the setup checklist and skill catalogue are system-org only 2026-10-11 04:01:46 +11:00
sam bf29d3640e fix(daemon): reown worktrees a killed container left root-owned 2026-10-11 04:01:19 +11:00
sam de15bffc2f fix(telegram): publish /approve, which the router already serves 2026-10-11 03:57:10 +11:00
sam 1377e7ebf2 feat(webui): add a member by provider subject or email
POST /api/orgs/{id}/members binds a person by subject, or by email as a
placeholder. The first sign-in whose provider vouches for that email
claims it: the State Store moves the binding onto the provider's subject
in one statement, so a second account with the same address finds
nothing to claim and later sign-ins rest on the subject alone.
2026-10-11 03:54:49 +11:00
sam d53ebbda95 fix(daemon): last poll reports the least recently polled loop 2026-10-11 03:53:11 +11:00
sam b227df3c3f feat(storepkg): widening updates wait for approval, rollback restores the previous pin 2026-10-11 03:48:31 +11:00
sam a3e0e5c1bb fix(daemon): bound a package workflow's tools by its accepted authority 2026-10-11 03:48:31 +11:00
sam 116f8a10fb style(webui): rewrap the assets comment 2026-10-11 03:48:17 +11:00
sam 116b03534f fix(staterpc): a task's own credential cannot replace its workflow pin
The pinned definition carries the package a run's authority is read
from. A task-scoped Update now reaches the store without a definition,
and UpdateTask keeps the stored pin when none is sent; the workflow name
still moves for handoffs and the daemon re-pins it.
2026-10-11 03:47:17 +11:00
sam b697a12b89 fix(webui): an unserved /api path is a 404, not the dashboard page 2026-10-11 03:46:40 +11:00
sam 7b64699ff1 refactor(workflow): delete routing overrides nothing applied 2026-10-11 03:43:25 +11:00
sam 1598b3cc74 docs(agents): extend the finish-and-shrink rule to 2026-10-20 2026-10-11 03:39:31 +11:00
sam 212742e880 feat(forgerpc): bound a package workflow's forge calls by its accepted authority 2026-10-11 03:37:16 +11:00
sam 6693e60f6f fix(access): an unloaded chain is a refusing engine, not nil
Webhook intake answers 503 and counts no refusal while the chain loads.
The readiness probe reads the chain on every check. The UI never exits on
a first-load failure. An invalid instance policy closes the instance but
lets an instance admin repair policies. Outage pages are 503, and the
health detail names the unloaded chain.
2026-10-10 22:19:02 +11:00
sam c097e50405 fix(archieui): warn when the dashboard is reachable off-host over plain HTTP
The dashboard serves plain HTTP, so a non-loopback bind carries the
session cookie and provider token in clear unless TLS terminates in front.
Warn at startup unless the operator declares a proxy with
[web].trust_forwarded_headers, and record the choice in deployments/README
where the dashboard is deployed. Point the non-loopback token error at
that surviving doc, not the deleted UI-boundary PRD.
2026-10-10 22:09:05 +11:00
sam ccbc9c7844 fix(webui): the wildcard listen address is not loopback
IsLoopback(":8484") returned true: the empty host fell through the
switch as loopback, so a wildcard bind skipped the non-loopback token
requirement and served every interface unauthenticated. An empty host is
the wildcard, not loopback.
2026-10-10 22:09:05 +11:00
sam 948a8d34fd test(worktree): give each transport its own branch
The four subtests share one bare repo but pushed the same feat/1-test
branch. Their commits have the same tree, message, author and parent, so
two commits in the same second hash identically and the second push is a
no-op -- passing by luck -- while a second of clock drift makes it a
non-fast-forward failure. One issue per subtest gives each its own branch.
2026-10-10 22:05:22 +11:00
sam e078b5c4bc fix(webui): refuse API requests until the access chain loads
A State Store that was unreachable at startup left the access chain nil,
and authorize treated nil as no chain at all: every request passed the
credential check, no principal or org reached the context, the State Store
defaulted each call to org-sys, and the instance-only surfaces were served.
The dashboard stayed open until a restart.

Build a chain with no engine when the first load is Unavailable; it refuses
every request with 503 (or the notice page for a browser) until the first
load succeeds, and Run keeps retrying so the dashboard recovers without a
restart. A denial is not recorded, because nothing was evaluated.
2026-10-10 21:55:02 +11:00
sam 4a24c0b31f fix(memory): each org's agent memory is its own 2026-10-10 21:20:37 +11:00
sam e988d3cc40 fix: close cross-org writes and move existing rows into their owner's org 2026-10-10 20:31:54 +11:00
sam cd20aeadad fix(webui): instance health, logs, config and identities are system-org only 2026-10-10 20:20:07 +11:00
sam 1bb7ba1248 fix(postgres): tasks, events, steps and run stats stay in their org 2026-10-10 20:18:29 +11:00
sam fdafec3d9d fix(postgres): event sources, types, mappings and bindings stay in their org 2026-10-10 20:14:23 +11:00
sam 0fcf792db3 feat(gateway): chat sessions belong to one org
Sessions carry an org; listing, reading and writing a session, its
messages and its turns are limited to the caller's org. Another org's
session reads as not found and cannot be overwritten or appended to.
Sessions from before orgs belong to the system org.
2026-10-10 18:57:26 +11:00
sam 9cd52dd9aa feat(archiegateway): chat resolves models in the caller's org and bills it
The State Store answers the calling principal's org aliases
(CallerModels) and bills chat usage to that org; usage records no longer
carry a caller-named org. Chat lists and resolves aliases for the
caller's org and fails rather than substitute one it may not use.
2026-10-10 18:53:57 +11:00
sam 5290dab199 feat(controlplane): per-org model access, org aliases and org providers
The system org sets each org's allowed models and whether it may add its
own providers (org-model-policy); an org with no entry may use nothing.
An org's aliases and providers are checked against that policy on save
and filtered by it at resolution. Tasks get their org's providers, keyed
by the org's credential binding through the egress proxy. Org Settings
gains a Models tab; instance Models gains Org access.
2026-10-10 18:48:40 +11:00
sam e17c7895e4 feat(daemon): resolve a task's models and credentials in the task's own org
The State Store answers a task's effective model aliases (instance aliases
with the task org's own over them), deriving the org from the task. Model
egress and the run credential resolver use the task's stored org instead
of the identity mapping in config.
2026-10-10 18:40:53 +11:00
sam e4ea2eaabc feat(statestore): record a usage row for every model call
Task agent runs and MCP sampling record against their task (billed to the
task's org); gateway chat turns, titles, sampling and curators record
against the system org. Rows carry source, workflow, step, alias,
provider/model and tokens for later chargeback.
2026-10-10 18:32:43 +11:00
sam 12b9a57404 feat(config): model aliases replace model roles, resolved by one resolver
Agent steps, implement, pr-review, MCP sampling, chat, embedding and
transcription resolve a model alias (empty means the purpose's default)
and fail on an unknown alias instead of falling back. The step editor and
chat pick aliases; workflow definitions naming an unknown alias are
refused on save.
2026-10-10 17:29:52 +11:00
sam 514c6cf61c fix(daemon): keep model provider keys out of agent containers via the egress proxy 2026-10-10 17:12:09 +11:00
sam bef0ecea3a feat(ui): workflow list shows running count and latest run result 2026-10-10 11:50:27 +11:00
sam 5bb5d9b660 feat(ui): quieter workflows page: hover data edges, apply chip, restore behind confirm 2026-10-10 11:49:32 +11:00
sam f30405ea2b fix(ui): input group addon no longer steals clicks from its select 2026-10-10 11:26:42 +11:00
sam f9efae8a5e fix(archiegateway): keep the chat model list non-nil when empty 2026-10-10 11:17:37 +11:00
sam 819cfc655b refactor: slices.Sorted over map keys and one shared IsNil check 2026-10-10 11:14:36 +11:00
sam 5afe5bcf99 refactor: use slices.Index/Delete and slices.Clone for small copies 2026-10-10 11:14:36 +11:00
sam 0c25b2f4a6 refactor(workflow): decodeSettings returns a typed value 2026-10-10 11:14:36 +11:00
sam 252895e01c refactor(staterpc): use mapValues for proto list conversions 2026-10-10 11:05:59 +11:00
sam 3ea49c7e07 refactor(gateway): drop duplicate string arg helpers 2026-10-10 11:05:59 +11:00
sam 66a27bd7ef refactor(controlplane): make layerResource decode into a typed value 2026-10-10 11:05:59 +11:00
sam b070ce4715 refactor(eda): use slices.Sorted over maps.Keys for ordered keys 2026-10-10 11:05:58 +11:00
sam 84b581fde6 refactor(webui): decode JSON request bodies with generic decodeBody 2026-10-10 11:05:58 +11:00
sam af5e11a734 refactor(domain): share one generic lifecycle registry for curators and memory engines 2026-10-10 11:05:58 +11:00
sam 522ced45ff feat(ui): lanes for parallel and switch steps on the workflow canvas 2026-10-10 10:52:52 +11:00
sam 79d7f2fa68 feat(workflow): switch step runs the case matching a reference 2026-10-10 10:52:52 +11:00
sam eb3ee14a0e feat(ui): step roles, numbers and run trace on workflow canvas 2026-10-10 10:37:47 +11:00
sam 6ed6b28d02 build(ui): refresh dashboard bundle 2026-10-09 20:06:01 +11:00
sam 2f4e9bbe14 Merge branch 'main' into codex/control-plane-4q28 2026-10-09 20:03:52 +11:00
sam 3057f054b2 style: remove space between icons and labels 2026-10-09 19:56:27 +11:00
sam a4e7ea8cc3 fix(kit): migrate runtime to sandbox kit m8 2026-10-09 18:31:56 +11:00
sam c812192800 feat(ui): author parallel workflow branches on canvas 2026-10-09 18:18:45 +11:00
sam 81121d03b0 docs(agents): omit status the maintainer can already see 2026-10-09 18:16:45 +11:00
sam e5f03587c9 docs(agents): post externally only when told 2026-10-09 18:09:11 +11:00
sam c25e70a588 docs(agents): act on clear direction and keep replies short 2026-10-09 18:04:38 +11:00
sam b97c79aa0b feat(ui): edit workflow call interfaces on the canvas 2026-10-09 18:04:08 +11:00
sam 2d2c39f9e5 fix(daemon): reconcile call waits from durable steps 2026-10-09 17:56:15 +11:00
sam 673a5bc2ce chore(release): v1.52.0
deploy / gate (push) Canceled after 0s
deploy / archied-image (push) Canceled after 0s
deploy / agent-image (push) Canceled after 0s
deploy / dist-zip (push) Canceled after 0s
2026-10-09 13:09:59 +11:00
sam df5ee243b1 feat(harness): create the setup PTY at the browser's size
The setup terminal created its PTY at a fixed 24x80 while the browser
measured its own box and threw the measurement away, so a full-screen CLI
login drew into a corner of a much larger xterm.

OpenRequest now carries a ConsoleSize with the target frame; the route
reads cols/rows from the WebSocket URL, falling back to 24x80 when they
are absent or out of range, and the daemon creates the PTY at that
geometry. The component sends the size it measured before connecting.

Live resize is the follow-on: it needs a control channel on the WebSocket,
which today is a raw byte copy.
2026-10-09 13:07:29 +11:00
sam 71fa670512 fix(harness): keep the setup container alive with archie's own binary
A setup session started its container with "sleep infinity" and opened the
PTY with "sh -l". Neither binary is guaranteed: the spec capability whose
floor is a shell and bash is com.docker.sandbox/sbx@1, and archie's
supported map does not list it, so a Kit image need not provide either. A
missing sleep killed the container at start and a missing sh failed the
exec, both as an opaque error.

The container now runs archie-agent hold as its entrypoint. The agent
binary is archie's own and already mounted into every Kit, so the
keepalive needs nothing the image may lack and the container env stays
empty. Before opening the PTY, harnesssession probes for bash then sh and
returns ErrNoShell naming the Kit reference when neither is present.

Test: TestPickShell covers bash-preferred, sh-only, neither, and a probe
that cannot run.
2026-10-09 13:05:05 +11:00
sam d91ab1ab0f fix(channels): normalize stored settings and deduplicate save errors 2026-10-09 13:02:23 +11:00
sam 1eb05f0d24 test(ui): prune brittle checks and document agent rules 2026-10-09 12:51:57 +11:00
sam a49e939609 fix(ui): clean dependencies and add task table sorting 2026-10-09 12:46:05 +11:00
sam e76e2fd317 chore(lint): pin the linter to 2.14.0
The setup task is the only installer, so the pin follows the version in
use rather than holding it back, and CI follows the pin.
2026-10-09 12:32:05 +11:00
sam 2418a81a9e fix(egress): use context-aware dial and request in the token-endpoint test
net.Dial and httptest.NewRequest fail the noctx linter, so the gate was red
on the commit that added the test.
2026-10-09 12:27:45 +11:00
sam c4a441d1c3 chore(task): install the pinned tooling with task setup
task setup installs the golangci-lint CI pins, the protobuf generators and
the dashboard dependencies, and warns by name about the tools it cannot
install (Go, Task, Node, Docker).

The linter version now lives in one file read by both the setup task and
both workflows, so a local checkout and the gate cannot run different
linters.
2026-10-09 12:27:45 +11:00
sam c5c145f325 test(archieui): restore the archie-ui import-boundary check
Only the daemon owns containers, and the dashboard consumes a harness
session over its gRPC contract. Go imports are atomic, so a single
reference relinks a runtime, and the boundary holds only while something
re-measures it. cmd/archie-ui/architecture_test.go enforced this until it
was purged with the test suite; nothing enforced it since.

The restored check runs go list -deps . and fails if the dashboard links
internal/container, internal/app/archied or internal/agentexec. It is
transitive, which is what "links" means, and runs in the existing test
step.
2026-10-09 11:42:39 +11:00
sam 3683ce9082 fix(egress): refuse a token endpoint the run is not granted
forward intercepted a declared OAuth credential's token endpoint only when
the run was granted the service. For an optional credential it fell
through to the reverse proxy, so the provider's real access and refresh
tokens reached the container un-sentinelized, contradicting the harness
contract's "the real credential never enters the container".

A matched token endpoint is now refused when the service is not granted
(403 for an optional credential; a required one keeps its existing 502).
The table test drives forward against a recording upstream and asserts
both the status and that the upstream was never contacted.
2026-10-09 11:42:39 +11:00
sam 80dc6ec8d0 fix(deps): pin sandbox-kit-spec to the API the code is written against
The dependency bump moved sandbox-kit-spec from v3.0.0-m.6 to m.8 without
adapting the code, so the tree no longer builds: CredentialCapability.Phase
is now spec.Phases, not a string (internal/infrastructure/egress), and
Client.Assemble became a package-level fetch.Assemble that takes a Loader
and a LayerValidator instead of spec.MergeOptions (internal/infrastructure/
kitrun).

Migrating to m.8 is not a version-string change: m.8 pushes kit integrity,
safe layer extraction and resource limits onto the caller, so it is
security-relevant work, not an unblock. Pin back to m.6, which the code is
written against, and leave the migration to a deliberate change.

Verified: go build ./..., go vet ./..., go test -short ./... all pass with
this pin. The gate's ui:check and lint steps fail for unrelated
toolchain reasons recorded separately.
2026-10-09 11:42:33 +11:00
sam 12d6585586 ```
chore: update ui dependencies
```
2026-10-09 11:27:48 +11:00
sam 53916b0f96 chore: bump Go dependencies and toolchain 2026-10-09 11:23:04 +11:00
sam badbbf84bd fix(config): normalize default org for harness credentials 2026-10-09 10:01:35 +11:00
sam d98352d003 feat(settings): classify startup and live runtime settings 2026-10-09 09:57:57 +11:00
sam 542bd88040 feat(logging): aggregate recent service logs in dashboard 2026-10-09 09:57:57 +11:00
sam 81384939a2 fix(gateway): attribute chat settings to signed-in users 2026-10-09 09:57:57 +11:00
sam c8ed5619c6 fix(webui): explain personal token availability 2026-10-09 09:57:57 +11:00
sam 3b79155602 refactor(webui): remove unused configuration schema 2026-10-09 09:57:57 +11:00
sam 552ee93a47 fix(settings): show configuration reload failures 2026-10-09 09:57:57 +11:00
sam 7bf429b599 fix(gateway): read task logs through state store 2026-10-09 09:57:57 +11:00
sam 0792f0ba84 fix(ui): render unknown routes and correct status description 2026-10-09 09:57:56 +11:00
sam 02ab747ca9 fix(configuration): reject unknown conf.d files 2026-10-09 09:57:56 +11:00
sam c53a98ee87 feat(curators): edit and apply definitions live 2026-10-09 09:57:56 +11:00
sam 1e4aa8badc feat(settings): edit profiles bindings and playbooks 2026-10-09 09:57:56 +11:00
sam 1e76c99b48 feat(settings): show resource consumers and apply status 2026-10-09 09:57:56 +11:00
sam ff9b6c7196 feat(presence): report workers and broker on status 2026-10-09 09:57:56 +11:00
sam 7680dc5079 feat(presence): expose service dependency checks 2026-10-09 09:57:56 +11:00
sam d0370ed9f7 fix(daemon): inherit live settings for identity tasks 2026-10-09 09:57:56 +11:00
sam 7e3b3dbd78 test: verify enablement reports channel running 2026-10-09 02:24:47 +11:00
sam 6b71dc8108 fix(workflow): enforce optional result booleans only when required 2026-10-09 02:22:19 +11:00
sam 307f2fc9e2 fix(messaging): refuse failed channels as enablement success 2026-10-09 02:22:19 +11:00
sam 455de3d799 fix(workflow): retry a transient read while awaiting a callee 2026-10-09 02:21:49 +11:00
sam 4e74ed37b4 fix(workflow): preserve a reference's type between steps 2026-10-09 02:10:43 +11:00
sam 7ef58293be feat(harness): serve a daemon-side setup session to the dashboard
The dashboard's setup terminal was a 503: it had a WebSocket route, an
authorizer and an xterm.js surface, but nothing owned a container. The
daemon owns containers, so the session contract lands there and the
dashboard consumes it over a new gRPC contract.

One session concept, two targets. A setup session is addressed by Kit
profile and is ephemeral; a run session (archie-core-qi91) is addressed by
task and attempt and is bounded by the task. This slice serves setup and
answers Unimplemented for run.

The contract, the authorisation, the lifetime and the egress phase are
settled in docs/prds/external-agent-harness.md, "Session contract". The
short version: the daemon serves harness.v1.HarnessService; the dashboard
authorizes the operator as an update on the secret and then dials the
daemon with the harness service token, required even on loopback because a
session is a shell holding an org's credential; the setup container runs
no worker, mounts no worktree, holds a keepalive, and carries an empty
container environment. Install hooks run at the install phase and the
session enters runtime before the shell, so a runtime-phase credential's
token endpoint is the one the proxy intercepts.

Capture is gated by the egress proxy on a granted service, and a setup
session has no task, so the daemon registers a session grant: the
session's proxy token resolves to the org's bindings for the Kit's
declared services. The real token never enters the container: the token
endpoint response is sentinelized and stored as the org's harness secret.

Every teardown path releases the pool slot, revokes the egress session and
the grant, removes the sandbox network, and removes the session's volumes.
The listener requires HARNESS_TOKEN and stops with a 10s grace before
Stop, so an open session cannot hold shutdown to the watchdog leash.
2026-10-09 01:38:41 +11:00
sam 27c9647a13 fix(container): reap orphaned processes in Kit containers 2026-10-09 01:37:57 +11:00
sam 89541eb3b0 chore(lint): drop nolint directives that suppress nothing
29 of 59 directives did nothing. Removing each one individually and
re-running the gate revealed no issue from any of them: 20 named gosec,
which is not enabled in .golangci.yml and never has been, plus 4 nilerr
in functions that return no error, 2 dupl where dupl reports nothing
repo-wide, and one each of contextcheck, nestif and unused. The funlen
name on settings.go:147 is gone for the same reason; its gocyclo and
cyclop names are both live.

The unused directive in webui/authorize.go sat on `var _ = slog.Default`,
a statement that only existed to keep the log/slog import alive. Both go.

The 30 remaining directives all suppress a real issue.
2026-10-08 23:29:19 +11:00
sam 3a3a714cc4 fix(modelcatalog): never serve a disabled provider from the ambient key
disableProvider already cleared a failed provider's credential, but that
made it indistinguishable from a provider nobody configured, so
usableProvider fell back to the catalog's ambient env var (e.g.
OPENAI_API_KEY) and sent turns with a key the operator did not choose.
Mark the provider Disabled when its key fails to resolve, refuse it in
usableProvider, and carry the flag through mergeProviders so a catalog
refresh cannot resurrect it.
2026-10-08 22:59:23 +11:00
sam b107e4452a fix(webui): keep the dashboard token out of the startup log
DashboardURL no longer takes a token, so the startup line cannot embed
?t=<token> in the printed URL. The log now names the token file instead.
2026-10-08 22:57:28 +11:00
sam 2b514cb526 feat(scheduling): refuse a job payload that names a dangerous command
A scheduled job runs unattended, so a payload that restarts the daemon,
formats a disk or powers the machine down is refused when the job is
created rather than when it fires. scheduling.RefuseDangerousJob is the
one rule every creation path reaches: the dashboard, any control-plane
client and the config seed all replace the schedules document through the
same validator, so a new token is added in one place.
2026-10-08 22:45:28 +11:00
sam 630c30de79 fix(archiemessaging): apply channel enablement live 2026-10-08 22:44:15 +11:00
sam a69b3455d3 refactor(tools): back generate_video with ai-sdk's minimax video provider
Delete the hand-rolled MiniMax HTTP/poll client; ai-sdk v0.1.43 already
ships provider/minimax implementing video.Provider with the same endpoints,
defaults and poll semantics. The tool now takes a video.Provider and the
composition root builds it, matching the transcription and embedding
capabilities.
2026-10-08 22:42:49 +11:00
sam ddf8383784 fix(archiemessaging): recover a channel that fails on its own 2026-10-08 22:36:09 +11:00
sam c1bbeba18f fix(archied): wait for in-flight tasks on a single-cycle run
A drain pass no longer waits for the tasks it submitted, so -once (the
systemd timer and test invocation) returned while their containers were
still running and the process then tore them down. Wait for the pass's
tasks here, restoring what the flag means.
2026-10-08 22:31:07 +11:00
sam 1f99fb592e fix(daemon): let a run waiting on a callee release its capacity
A wait:true workflow.call run held its global slot and its repo's place for
the callee's whole life, and the drain pass waited for its own tasks, so a
callee queued by that pass was never claimed. With containers.max_concurrency
saturated, or a repo that does not allow concurrency, the wait could not end.

A drain pass now returns once the stream and the claim fallback are empty, and
shutdown waits for in-flight work once. Suspend/Resume give a run's slot and
repo place back while it waits on a callee, driven by the workflow call events
on the daemon's bus; a hold records one slot and one place, so a second
re-acquire gives back what it took.
2026-10-08 22:31:07 +11:00
sam aae01431b4 refactor(workflow): keep the call key change inside the lint gate
golangci-lint failed on the change and had not before: gofumpt wants
StartCall's same-typed parameters grouped, and runWorkflowCall reached a
cyclomatic complexity of 16. The unwaited call's early close moves into
closeUnwaitedCall on *TaskContext -- it needs only the call step, the
callee and the settings -- which leaves runWorkflowCall below the limit
with its behaviour unchanged.

The walk's id-less call case is deleted: ParseDefinition now refuses that
definition before validateWorkflowCalls runs, so it passed for the wrong
reason. The refusal is pinned where it happens instead, by the compile
case in TestWorkflowCallKeyIsTheDeclaredPath, now asserting the whole
message so it still names the step it refused.
2026-10-08 22:31:07 +11:00
sam 3c99f69fee fix(workflow): key a call on its callee, and free a terminal child's key
A call's key was the step's declared path alone, so one task running a
second definition at a colliding path -- workflow.handoff, human.approve
and a definition re-pin all change it -- was handed the first callee's
child, waited on the wrong workflow and published its outputs. The key
now carries the callee too, while the step row and the call's events stay
named and staged for the call site, so a rail pairing an event with its
step still attaches them.

A child that reached dead or closed_wont_do can never do the work and
holds no result the caller can use, so it leaves the call key's index and
frees the key for the caller's retry to start a fresh child. parked and
rejected keep theirs: a parked child is recoverable by retrying it, and a
rejected child ran.

The id a call step must declare is now refused where the definition is
compiled, which is also the path a pinned definition takes, so an id that
spells a step type is accepted again and no stage needs a heuristic.
2026-10-08 22:31:07 +11:00
sam b7fd13923a fix(workflow): key a workflow.call on its declared step path
A call site had no durable identity, so any retry of it -- a lost
EnqueueCallTask reply, a failed call-step record, a requeued attempt that
re-runs the stage -- started a second child and its work happened twice.

The call step's declared path within the run is that identity: wrapStep
sets it (branch-qualified, restored on return), a workflow.call step must
declare an id so two call sites cannot share one, and the key crosses
EnqueueCallTaskRequest.call_key into tasks.call_key, which EnqueueCallTask
upserts on so the retry returns the child already started.
2026-10-08 22:31:07 +11:00
sam 0a47a95cc3 fix(agentexec): enforce capture tool result schemas 2026-10-08 22:30:12 +11:00
sam ef8d9af469 fix(workflow): apply step retry and on_failure to command.run 2026-10-08 22:22:13 +11:00
sam 549f36a1f2 fix(ui): stop the command palette scrolling sideways 2026-10-08 21:58:22 +11:00
sam 187e6e81ee fix(ui): anchor the command palette and stop scroll stealing its highlight 2026-10-08 21:57:20 +11:00
sam e60ad551f0 fix(webhook): authorize delivery against the source's org 2026-10-08 21:43:28 +11:00
sam a0617ad437 test(postgres): derive the org fixture source from its insert 2026-10-08 21:40:24 +11:00
sam 411736fdaa fix(captures): keep count-retention inside the inserted capture's org 2026-10-08 21:27:54 +11:00
sam bea6567933 fix(captures): scope the capture list to the acting org 2026-10-08 21:26:07 +11:00
sam d6b693a9b2 fix(sources): store the acting org on source creation 2026-10-08 21:25:49 +11:00
sam b8bedf4d7b fix(capture): inherit source org on ingestion 2026-10-08 21:02:21 +11:00
sam 5c53dfb91b fix(workflow): resolve definitions in the owning org 2026-10-08 20:41:08 +11:00
sam e0cbbc0895 chore(release): v1.51.1
deploy / gate (push) Canceled after 0s
deploy / archied-image (push) Canceled after 0s
deploy / agent-image (push) Canceled after 0s
deploy / dist-zip (push) Canceled after 0s
2026-10-08 01:42:29 +11:00
sam 40dd546564 fix(release): a dependency bump is releasable for every binary 2026-10-08 01:42:27 +11:00
sam b89d8cbab2 fix(deps): ai-sdk v0.1.43 sends gpt-6 tool calls through the Responses API 2026-10-08 01:41:44 +11:00
sam 4b62912343 chore(release): v1.51.0
deploy / gate (push) Canceled after 0s
deploy / archied-image (push) Canceled after 0s
deploy / agent-image (push) Canceled after 0s
deploy / dist-zip (push) Canceled after 0s
2026-10-08 01:37:28 +11:00
sam 8a8c8cf7e9 feat(chat): the dashboard chat records voice, downloads files a turn sends, and offers /model and /personality choices 2026-10-08 01:37:21 +11:00
sam 201529414b fix(gateway): providers resolve again when a secret engine starts after boot 2026-10-08 01:34:55 +11:00
sam 3da1f918a4 chore(release): v1.50.1
deploy / gate (push) Canceled after 0s
deploy / archied-image (push) Canceled after 0s
deploy / agent-image (push) Canceled after 0s
deploy / dist-zip (push) Canceled after 0s
2026-10-08 00:40:03 +11:00
sam f8d8e5b1e2 fix(update): the watchdog logs the health report it rolled back on 2026-10-08 00:39:57 +11:00
sam 826d4849df chore(release): v1.50.0
deploy / archied-image (push) Canceled after 0s
deploy / agent-image (push) Canceled after 0s
deploy / dist-zip (push) Canceled after 0s
deploy / gate (push) Canceled after 33s
2026-10-08 00:25:35 +11:00
sam 0a18bfe21b feat(chat): the dashboard chat answers tool approvals and questions, and /update offers Install 2026-10-08 00:25:23 +11:00
sam 5976b94fbc fix(gatewayrpc): carry tool approvals and ask_user questions between a channel and the gateway turn 2026-10-08 00:22:37 +11:00
sam 698548d6ef chore(release): v1.49.4
deploy / gate (push) Canceled after 0s
deploy / archied-image (push) Canceled after 0s
deploy / agent-image (push) Canceled after 0s
deploy / dist-zip (push) Canceled after 0s
2026-10-07 23:58:52 +11:00
sam d0584279e5 fix(ui): tabbed Status and Extensions, one Add MCP server, repos page renders without repos, save bar rests at the viewport edge 2026-10-07 23:46:34 +11:00
sam 62f248fd2c fix(ui): the Versions table's status column ends at the card's edge 2026-10-07 23:35:12 +11:00
sam 8f2750a266 chore(release): v1.49.3
deploy / gate (push) Canceled after 0s
deploy / archied-image (push) Canceled after 0s
deploy / agent-image (push) Canceled after 0s
deploy / dist-zip (push) Canceled after 0s
2026-10-07 23:34:21 +11:00
sam 2ea7f7cd38 fix(ui): the release channel control sits at the Updates card's edge, not mid-row 2026-10-07 23:34:13 +11:00
sam 75b3810856 fix(archieui): the Versions card compares against the running daemon, so an available update can be installed 2026-10-07 23:29:03 +11:00
sam 30419532ba chore(release): v1.49.2
deploy / archied-image (push) Canceled after 0s
deploy / agent-image (push) Canceled after 0s
deploy / dist-zip (push) Canceled after 0s
deploy / gate (push) Canceled after 1m51s
2026-10-07 23:18:11 +11:00
sam fff67838e0 fix(releaseupdate): read GitHub releases directly; drop go-selfupdate and its openpgp dependency 2026-10-07 23:18:03 +11:00
sam 15ebe24927 chore(release): v1.49.1
deploy / gate (push) Failing after 3m43s
deploy / archied-image (push) Skipped
deploy / agent-image (push) Skipped
deploy / dist-zip (push) Skipped
2026-10-07 23:10:38 +11:00
sam b2c1629925 fix(ui): an expired session reloads to the sign-in page instead of showing a banner 2026-10-07 23:05:52 +11:00
sam 0e8fcf70a0 chore(release): v1.49.0
deploy / gate (push) Failing after 5m18s
deploy / archied-image (push) Skipped
deploy / agent-image (push) Skipped
deploy / dist-zip (push) Skipped
2026-10-07 22:17:00 +11:00
sam 4673b435b7 fix(extension): confine egress only for packages that declare egress hosts 2026-10-07 22:15:06 +11:00
sam 1e062fd7f0 fix(releaseupdate): reuse a found release for ten minutes to stay under GitHub's unauthenticated rate limit 2026-10-07 01:17:21 +11:00
sam 74a3ec925c feat(releaseupdate): archied update check finds releases through go-selfupdate; archie-update-check is deleted 2026-10-07 01:12:11 +11:00
sam 4944ec8473 chore(release): v1.48.1
deploy / gate (push) Failing after 3m28s
deploy / archied-image (push) Skipped
deploy / agent-image (push) Skipped
deploy / dist-zip (push) Skipped
2026-10-07 01:00:30 +11:00
sam 204b65a0e3 fix(controlplane): update-settings selects a release channel; services run only the installed update scripts 2026-10-07 01:00:28 +11:00
sam cbedb8a504 chore(release): v1.48.0
deploy / gate (push) Failing after 5m4s
deploy / archied-image (push) Skipped
deploy / agent-image (push) Skipped
deploy / dist-zip (push) Skipped
2026-10-07 00:57:07 +11:00
sam a8dbab0cff feat(controlplane): update commands are an update-settings kind edited on Status, defaulting to the installed scripts 2026-10-07 00:57:01 +11:00
sam 789ff73fe9 fix(scripts): archie-update-install loads the archie env file so a shell run can back up the bundled database 2026-10-07 00:53:23 +11:00
sam 5176938e95 ci(deploy): build the archied and agent images in parallel so the release zip waits only on the agent image 2026-10-07 00:35:13 +11:00
sam 69a1ee677d chore(release): v1.47.3
deploy / gate (push) Failing after 3m29s
deploy / build-and-push (push) Skipped
deploy / dist-zip (push) Skipped
2026-10-07 00:33:23 +11:00
sam b3cf2c59a8 fix(webui): extension Accept and Remove dialogs act on the selected extension 2026-10-07 00:32:44 +11:00
890 changed files with 45849 additions and 22607 deletions

No files matched your search

+13 -11
View File
@@ -54,15 +54,15 @@
# Disabled by default; enable only when an integration needs fresh .beads/issues.jsonl.
# Use relative paths under .beads/ for JSONL import/export filenames.
export:
auto: false
path: issues.jsonl
interval: 60s
# issues.jsonl/interactions.jsonl are .gitignore'd local-only files --
# Dolt (synced via sync.remote) is the source of truth. Leave auto off
# and git-add off; do not re-track these in git.
git-add: false
auto: false
path: issues.jsonl
interval: 60s
# issues.jsonl/interactions.jsonl are .gitignore'd local-only files --
# Dolt (synced via sync.remote) is the source of truth. Leave auto off
# and git-add off; do not re-track these in git.
git-add: false
import:
path: issues.jsonl
path: issues.jsonl
# Integration settings (access with 'bd config get/set')
# Non-secret keys (stored in the database):
# - jira.url, jira.project
@@ -74,7 +74,9 @@ import:
# - github.token → use GITHUB_TOKEN env var instead
sync.remote: "git+https://github.com/samcharles93/archie-core.git"
dolt:
shared-server: false
auto-commit: "on"
shared-server: false
auto-commit: "on"
audit:
enabled: true
enabled: true
agent:
profile: team-maintainer
+57 -33
View File
@@ -30,7 +30,7 @@ permissions:
jobs:
gate:
runs-on: ubuntu-latest
runs-on: self-hosted
outputs:
archied_version: ${{ steps.release_tag.outputs.archied_version }}
steps:
@@ -66,29 +66,13 @@ jobs:
- name: Install quality tools
run: |
go install github.com/golangci/golangci-lint/v2/cmd/golangci-lint@v2.13.2
go install github.com/golangci/golangci-lint/v2/cmd/golangci-lint@$(cat .golangci-version)
go install mvdan.cc/gofumpt@v0.11.0
go install github.com/go-task/task/v3/cmd/task@v3.48.0
- name: Install protobuf tools
run: bash tools/proto.sh install
# The backup/restore tests shell out to pg_dump/pg_restore against a
# postgres:18 testcontainer; the runner's stock client is older and
# refuses the server version.
- name: Install PostgreSQL 18 client
run: |
sudo install -d /usr/share/postgresql-common/pgdg
sudo curl --fail --silent --show-error --location \
https://www.postgresql.org/media/keys/ACCC4CF8.asc \
--output /usr/share/postgresql-common/pgdg/apt.postgresql.org.asc
. /etc/os-release
echo "deb [signed-by=/usr/share/postgresql-common/pgdg/apt.postgresql.org.asc] https://apt.postgresql.org/pub/repos/apt ${VERSION_CODENAME}-pgdg main" \
| sudo tee /etc/apt/sources.list.d/pgdg.list
sudo apt-get update
sudo apt-get install -y --no-install-recommends postgresql-client-18
echo "/usr/lib/postgresql/18/bin" >> $GITHUB_PATH
- name: Ensure contract baseline ref
run: |
if ! git show-ref --verify --quiet refs/heads/main; then
@@ -101,7 +85,17 @@ jobs:
# Extensions run in an unprivileged user and network namespace, which
# Ubuntu 24.04 runners forbid through AppArmor by default.
- name: Allow unprivileged user namespaces
run: sudo sysctl -w kernel.apparmor_restrict_unprivileged_userns=0
run: |
if [ -e /proc/sys/kernel/apparmor_restrict_unprivileged_userns ]; then
sudo sysctl -w kernel.apparmor_restrict_unprivileged_userns=0
fi
unshare --user --map-root-user --net true
# The gate's uncached `go test ./...` runs the Docker-backed tests that
# -short skips, and those start a container from alpine:3. A fresh runner
# has no such image, so the suite failed on "No such image: alpine:3".
- name: Pull test container images
run: docker pull alpine:3
- name: Quality gate
run: |
@@ -170,11 +164,9 @@ jobs:
echo "This commit is not a release, so the images are published as :edge and sha-<sha>; :latest keeps naming the newest release."
fi
build-and-push:
outputs:
agent_digest: ${{ steps.agent.outputs.digest }}
archied-image:
needs: gate
runs-on: ubuntu-latest
runs-on: self-hosted
steps:
- name: Checkout
uses: actions/checkout@v4
@@ -264,6 +256,43 @@ jobs:
cache-from: type=gha,scope=archied
cache-to: type=gha,mode=max,scope=archied
agent-image:
outputs:
agent_digest: ${{ steps.agent.outputs.digest }}
needs: gate
runs-on: self-hosted
steps:
- name: Checkout
uses: actions/checkout@v4
with:
fetch-depth: 0
# Annotated tags: dereference to the commit, as archied-image does.
- name: Resolve the commit this ref points at
id: commit
run: |
sha="$(git rev-parse "${GITHUB_REF}^{commit}")"
echo "sha=$sha" >> "$GITHUB_OUTPUT"
echo "short=${sha:0:12}" >> "$GITHUB_OUTPUT"
- name: Set up Docker Buildx
uses: docker/setup-buildx-action@v3
- name: Log in to GHCR
run: echo "${{ secrets.GITHUB_TOKEN }}" | docker login ghcr.io -u "${{ github.actor }}" --password-stdin
- name: Resolve runtime version
id: versions
run: |
RUNTIME_TAG="$(git tag --points-at HEAD --list 'archie/v*' | sort -V | tail -1)"
RELEASE_TAG="$(git tag --points-at HEAD --list 'v*' | sort -V | tail -1)"
RUNTIME_VERSION="${RUNTIME_TAG#archie/}"
[ -n "$RUNTIME_VERSION" ] || RUNTIME_VERSION="${RELEASE_TAG#v}"
echo "runtime_tag=$RUNTIME_TAG" >> "$GITHUB_OUTPUT"
echo "release_tag=$RELEASE_TAG" >> "$GITHUB_OUTPUT"
echo "runtime=$RUNTIME_VERSION" >> "$GITHUB_OUTPUT"
- name: Build and push archie-agent
id: agent
if: steps.versions.outputs.runtime_tag != '' || steps.versions.outputs.release_tag != ''
@@ -283,14 +312,10 @@ jobs:
# Packages every host-run binary (archied, archie-gateway,
# archie-state-store, archie-ui, archie-messaging) as a
# linux/amd64 zip for a
# non-Docker install. Independent of build-and-push: it never touches the
# archied/archie-agent image jobs, and a gateway-only release (no
# archie/v* tag) still produces this zip -- it's keyed to the archied tag
# alone. archie-agent is deliberately excluded: it only ever runs inside
# the Docker image build-and-push already publishes, never as a host
# binary (archie-core-8cda.4.7.1).
# non-Docker install. It waits only for the agent image, whose digest
# release.json pins; archie-agent itself never ships as a host binary.
dist-zip:
needs: [gate, build-and-push]
needs: [gate, agent-image]
# Not gated on github.ref. A release is pushed as main plus both tags at
# one SHA, the concurrency group below collapses those into a single run,
# and the survivor is whichever ref happened to start last -- for
@@ -299,7 +324,7 @@ jobs:
# the ref that won the race, and skip only when there is genuinely no
# archied tag at HEAD.
if: needs.gate.outputs.archied_version != ''
runs-on: ubuntu-latest
runs-on: self-hosted
# Least privilege: the workflow default is contents:read, and only this
# job needs to write, to create the Release and attach the zip to it.
permissions:
@@ -351,8 +376,7 @@ jobs:
cp "dist/$cmd" "$name/"
done
cp deployments/INSTRUCTIONS.md CHANGELOG.md "$name/"
cp scripts/archie-update-* "$name/"
digest='${{ needs.build-and-push.outputs.agent_digest }}'
digest='${{ needs.agent-image.outputs.agent_digest }}'
test -n "$digest"
jq --arg image "ghcr.io/samcharles93/archie-agent@$digest" '. + {agent_image: $image}' release.json > "$name/release.json"
# README.md doesn't exist yet in this repo; include it once it does
+10 -24
View File
@@ -14,12 +14,19 @@ concurrency:
jobs:
check:
runs-on: ubuntu-latest
runs-on: self-hosted
steps:
- uses: actions/checkout@v4
with:
fetch-depth: 0
- name: Verify extension namespace support
run: |
if [ -e /proc/sys/kernel/apparmor_restrict_unprivileged_userns ]; then
sudo sysctl -w kernel.apparmor_restrict_unprivileged_userns=0
fi
unshare --user --map-root-user --net true
- uses: actions/setup-go@v5
with:
go-version-file: go.mod
@@ -33,28 +40,12 @@ jobs:
- name: Install quality tools
run: |
go install github.com/go-task/task/v3/cmd/task@v3.48.0
go install github.com/golangci/golangci-lint/v2/cmd/golangci-lint@v2.13.2
go install github.com/golangci/golangci-lint/v2/cmd/golangci-lint@$(cat .golangci-version)
go install mvdan.cc/gofumpt@v0.11.0
bash tools/proto.sh install
sudo apt-get update
sudo apt-get install -y --no-install-recommends ripgrep
sudo install -d /usr/share/postgresql-common/pgdg
sudo curl --fail --silent --show-error --location \
https://www.postgresql.org/media/keys/ACCC4CF8.asc \
--output /usr/share/postgresql-common/pgdg/apt.postgresql.org.asc
. /etc/os-release
echo "deb [signed-by=/usr/share/postgresql-common/pgdg/apt.postgresql.org.asc] https://apt.postgresql.org/pub/repos/apt ${VERSION_CODENAME}-pgdg main" \
| sudo tee /etc/apt/sources.list.d/pgdg.list
sudo apt-get update
sudo apt-get install -y --no-install-recommends postgresql-client-18
echo "/usr/lib/postgresql/18/bin" >> $GITHUB_PATH
pg_dump --version
pg_restore --version
- name: Ensure contract baseline ref
run: |
@@ -62,12 +53,7 @@ jobs:
git branch main origin/main
fi
# Extensions run in an unprivileged user and network namespace, which
# Ubuntu 24.04 runners forbid through AppArmor by default.
- name: Allow unprivileged user namespaces
run: sudo sysctl -w kernel.apparmor_restrict_unprivileged_userns=0
- name: Run definitive gate
run: |
task check
task --concurrency 1 check
git diff --exit-code
+1
View File
@@ -0,0 +1 @@
v2.14.0
+166
View File
@@ -5,6 +5,172 @@ release's per-component sections are labelled beneath its heading.
## [Unreleased]
## [1.54.0] - 2026-10-16
**Upgrade note:** services now authenticate each other with TLS certificates from your own CA instead of shared tokens. Before upgrading a host whose services listen on anything but loopback (any host running agent containers), issue a certificate per service (`archied`, `archie-gateway`, `archie-messaging`, `archie-state-store`, `archie-ui`, each with its name as the DNS SAN and both server and client auth), add `[tls] ca_file` and `cert_dir` to `config.toml`, and remove every `target_token`, the `*_TOKEN` service variables and the `--token`/`--*-token` flags, including archie-ui's `-token` and `-token-file`. A unit that still passes a removed flag fails to start. See `config.example.toml`.
### archied
- Services authenticate each other with mutual TLS. Each process presents `<cert_dir>/<service>.pem`, peers must chain to `ca_file`, and a caller is identified by its certificate rather than by what it claims. Renewed certificates are picked up without a restart. Without `[tls]`, services listen on and dial loopback only. The harness setup terminal requires `[tls]`.
- Every gRPC service caps the streams a connection holds, times out stalled handshakes, releases a dead client's streams, and drains for at most 10 seconds on shutdown. Messaging exits when its listener fails instead of staying up without delivering.
- The dashboard has user accounts. First run opens browser setup for the organisation, workspace and first administrator; people then sign in with a password, a passkey or your identity provider, and account administration lives under Access. The shared dashboard token is gone; reach the dashboard remotely through an HTTPS reverse proxy.
- `archied update install` updates a host in one transaction: it verifies the release, stages every binary, pulls the agent image by digest and backs up the database, then restarts the services in order and rolls everything back if archied is not healthy for two minutes.
- Each workflow runs as the identity it declares, checked on every attempt along with the identity that started it. A workflow that declares none is refused, a called workflow runs as its caller's declared identity, and shipped workflows run as the instance root.
- Test-run a workflow draft before saving it, with forge and push writes suppressed and recorded per step. The History tab lists a workflow's revisions, diffs them as YAML and on the canvas, restores one into the editor, and shows which revision each run used.
- Start a task from the dashboard. The dashboard home shows activity and setup, and a running task has a read-only Sandbox tab that follows its container.
- The topbar's Jump to is a command palette, settings lists open as accordions that wait for a new item to be saved, and history restores land as a draft applied from the save bar.
- Labels, hints, alerts and empty states say what is wrong and what to do, without tutorial or marketing prose.
- Org isolation is tighter: harness secrets, identity writes, principals, playbook dispatch and model selections in chat stay in the caller's org, instance settings are written by the system org only, and creating a source or reloading a channel is authorized against the target.
- Binding dispatch claims its capture and enqueues the task atomically, so a capture cannot start two runs, and the run-time workflow check sees the binding event's signature and address.
- Service-unhealthy alerts count their grace from the last report and also cover a service that went silent, and a restart no longer clears an alert still inside its grace. An invalid stored access policy raises an alert until it validates.
- A failed cancel of a called workflow no longer leaves the caller running, retention spares captures still waiting to be dispatched, and EDA playbooks load and report at start instead of on first dispatch.
- Chat model selection works again for provider models.
### archie-agent
- Each run acts as its workflow's declared identity.
- A task snapshot carries only the model limits its aliases resolve to.
- A build step's "no changes" result comes from its writable stages and clears once a commit lands.
- Test runs suppress forge and push writes and record what they would have done.
- The default persona is a reporting-focused identity prompt, without the one-question limit in chat.
## [1.53.0] - 2026-10-11
### archied
- Every organisation's data is its own. Chats, tasks, events, transitions, steps, tool calls, run statistics, event sources, event types, mappings, bindings and agent memory are read and written in the org that owns them, and a request can no longer write into another org. Rows written before orgs existed move to their owner's org on upgrade.
- An instance admin decides what each org may reach: the models it can use, and whether it may register providers under its own credentials. Org Settings gains a Models tab and instance Models gains Org access. The system org is never restricted; another org uses nothing until it is granted.
- Models are addressed by alias. Aliases replace model roles, one resolver serves every purpose (agent steps, implement, PR review, MCP sampling, chat, embedding, transcription), and an unknown alias fails instead of falling back to a default. The step editor and the chat picker offer aliases, and a workflow naming one that does not exist is refused when saved.
- Chat sessions belong to one org; a chat resolves its models in the caller's org and is billed to it, and a task run from chat posts its outcome back to that chat.
- Add a member by provider subject, or by email as a placeholder that the first sign-in the provider vouches for claims.
- The workflow canvas authors parallel branches as lanes, adds a switch step that runs the case matching a reference, and edits a call's interface. A step shows its role, its number and its run trace.
- The workflows list shows each workflow's running count and latest run result, the canvas reveals its data-edge handles on hover, and restoring a version asks for confirmation.
- Store packages can auto-update: a package given the policy is moved to the catalogue's newest digest by a State Store ticker. An update asking for more authority than was accepted waits as a pending update with the running version untouched; approving it accepts the new authority, and a rollback restores the previous version with the authority accepted against it.
- Model provider keys never enter an agent container: the runtime reaches its providers through the egress proxy, which holds the keys. A task's own credential can no longer replace the workflow definition its run is pinned to, and a package workflow's triggers, tools and forge calls are bounded by the authority accepted for the package.
- The dashboard warns when it is reachable from off-host over plain HTTP, and the wildcard listen address counts as off-host rather than loopback. Instance health, logs, config and identities, the setup checklist and the skill catalogue are system-org only, and API requests are refused until the access chain has loaded.
- `/approve` reaches Telegram; the router served it but nothing published it.
- A worktree a killed container left root-owned is taken back by the daemon, and a PR review's benchmark runner works again on configured models and public PRs.
- An unserved `/api` path returns 404 instead of the dashboard page, and an input group's select no longer loses clicks to its addon.
### archie-agent
- The runtime reaches model providers through the egress proxy, so provider keys stay out of the container.
- Agent steps resolve model aliases and their own org's models and credentials, failing on an unknown alias instead of falling back.
- A switch step runs the case matching its reference.
- A call waiting on another step is reconciled from durable steps, so it resumes after a daemon restart rather than stranding.
## [1.52.0] - 2026-10-09
### archied
- The harness page's setup terminal opens a real session: it starts an ephemeral container from a Kit profile, runs the CLI's own login and captures the OAuth tokens at the provider's token endpoint. Point the dashboard at the daemon with `[services.harness].target` and a `HARNESS_TOKEN` and it stops answering 503; the terminal is created at your browser's size. Takeover of a live task is not in this release.
- A credential's token endpoint is refused when the run is not granted that service, instead of proxy-ing the provider's real tokens into the container.
- Settings → Profiles, Credential bindings and Playbooks are editable from the dashboard, and curator definitions can be changed and applied without a restart.
- Settings say whether each one applies live or needs a restart, show what is consuming resources, and report whether an apply landed.
- Status shows each service's dependency checks, the agent workers and the broker.
- The Logs page reads recent logs from every service, not just the daemon.
- Captures and webhook sources are scoped to the org that owns them: the list, retention and delivery authorization all act in the right org, and a workflow definition resolves in its own org.
- Chat settings are attributed to the signed-in user, and Settings → Tokens explains when a personal token cannot be created.
- A channel enabled after Archie started works immediately; a channel that fails on its own recovers; a failed channel is no longer reported as enabled, and stored channel settings no longer save duplicate or unnormalized values.
- The task table sorts, the command palette no longer scrolls sideways or loses its highlight, unknown routes render a not-found page, and configuration reload failures are shown instead of passing silently.
- A scheduled job whose payload names a dangerous command is refused.
- A disabled model provider is never served from the ambient API key.
- `generate_video` is backed by ai-sdk's MiniMax video provider.
- A dashboard token is no longer written to the startup log.
### archie-agent
- `workflow.call`: a call is keyed on its declared step path, its key is freed once the callee reaches a terminal state, and a caller waiting on a callee releases its capacity instead of holding it.
- `command.run` honours step retry and `on_failure`.
- A value passed between steps keeps its declared type, and optional result booleans are enforced only when the workflow declares them.
- Capture tools have their results checked against the declared schema.
- A single-cycle run waits for its in-flight tasks before exiting, and identity tasks inherit live settings.
- Kit containers run Docker's init as PID 1, so orphaned processes are reaped instead of accumulating inside a long task.
## [1.51.1] - 2026-10-08
### archied
- Chat with OpenAI's gpt-6 models (gpt-6-luna and the rest) works with tools; a turn failed with "Function tools with reasoning_effort are not supported".
## [1.51.0] - 2026-10-08
### archied
- A secret engine enabled after Archie started (Bitwarden, for example) now supplies its keys straight away; providers that depended on it no longer stay disabled until a restart.
- The dashboard chat has a microphone button: record a voice message and it is transcribed like a Telegram voice note. It shows only when a `transcription` model role is set under Settings → Models.
- A file Archie sends in the dashboard chat is a download link for an hour, instead of a note that it can't be delivered.
- Typing `/model` or `/personality` on its own in the dashboard chat offers the choices as buttons.
## [1.50.1] - 2026-10-08
### archied
- When an update is rolled back, the log names the health check that failed instead of only "unhealthy".
## [1.50.0] - 2026-10-08
### archied
- Tools that need your approval ask for it again, and Archie can ask you a question or offer choices mid-reply. Since the services were split these never reached Telegram or the dashboard: approval-gated tools failed and questions were never asked.
- The dashboard chat shows approvals as Approve / Always this session / Deny, choices as buttons, and questions with a reply field.
- `/update` in the dashboard chat lists what's available with Install and Not now.
- Settings → Status → Versions installs the update it shows; its Install button sent an empty request.
## [1.49.4] - 2026-10-07
### archied
- Settings → Status and Settings → Extensions are split into tabs instead of one long page; the Versions table's status column ends at the card's edge.
- Tools & MCP has one "Add MCP server" button; the transport is chosen in the form, and an unnamed new server is not an unsaved change.
- Settings → Repositories shows the add field when no repositories are configured, instead of a blank page.
- The unsaved-changes bar always rests at the bottom of the window.
## [1.49.3] - 2026-10-07
### archied
- Settings → Status → Versions shows the Gateway as "Update available" with an Install button when a newer release exists; it read "Unknown" with no way to install. The Runtime row stays "Unknown" until an agent task has run.
- The release channel buttons sit at the Updates card's right edge instead of mid-row.
## [1.49.2] - 2026-10-07
### archied
- Carries the 1.49.0 and 1.49.1 changes, whose builds were never published. The update check reads GitHub releases directly and still offers only releases with a verifiable linux-amd64 zip.
## [1.49.1] - 2026-10-07
### archied
- When the dashboard session expires, the page returns to the sign-in screen instead of showing errors on every page until a manual refresh.
## [1.49.0] - 2026-10-07
### archied
- Extensions that declare no egress hosts (vault, email, gitea) reach the network again; only packages that declare their hosts are confined to them. The Accept dialog shows "network: unrestricted" for the rest.
- The update check finds releases itself, and reuses a found release for ten minutes so repeated checks stay under GitHub's rate limit.
## [1.48.1] - 2026-10-07
### archied
- **Settings → Status → Updates** now chooses only the release channel (stable, prereleases, or a pinned version). Update checks and installs always run the `archie-update-*` scripts installed beside the binaries; the dashboard can no longer set the command that runs.
## [1.48.0] - 2026-10-07
### archied
- **Settings → Status → Updates** edits the update check and install commands. They default to the `archie-update-*` scripts installed beside the binaries, so a zip install shows **Install** on the Versions card with nothing configured. The `update_check_command` and `update_install_command` keys under `[chat.telegram]` are no longer read; remove them from `config.toml`.
- `archie-update-install` run from a shell now reads `~/.config/archie/env`, so the pre-update backup of the bundled database no longer fails with `no password supplied`. Run it with `--auto` to install the newest release.
## [1.47.3] - 2026-10-07
### archied
- **Settings → Extensions**: **Accept** in the authority review now accepts the extension, and **Remove** removes it. Both previously closed the dialog without doing anything.
## [1.47.2] - 2026-10-06
### archied
+17 -2
View File
@@ -11,7 +11,7 @@ agent or tool in an isolated container, and the result lands somewhere for
review or action. A repository is an optional input to a workflow, not the point
of the product.
<temporary - this statement must be removed on 2026-10-10"> The product is
<temporary - this statement must be removed on 2026-10-20"> The product is
unfinished. Many features are half-built, and the codebase carries far more
code, binaries, docs and tooling than it delivers. The job right now is to
finish features and shrink the codebase, not to extend it. </temporary>
@@ -27,6 +27,20 @@ When sources disagree, the higher one wins:
5. Everything else (memories, comments, old docs) is a hint. Verify it against
the code before acting on it.
## Working with the maintainer
- **A clear direction is the go-ahead.** Do it in the same turn. Do not restate
the plan, offer options, or ask "shall I proceed?". Stop only for destructive
or outward-facing actions (deleting data, pushing, touching a live host).
- **Post outside the repo only when told.** Chat, tickets and any external
service get a message only on the maintainer's instruction.
- **Report, don't narrate.** Final replies say what changed, what was verified,
and anything blocked, in a few lines. No preamble, no recap of the request, no
closing offers of more help. Leave out what the maintainer can see or infer:
a clean tree, unpushed commits, a passing gate (commits imply it), and steps
skipped that the task never asked for. Name a skipped step only when the
task required it.
## How to decide
- **Broken beats new.** A bug that blocks work gets fixed now, with the smallest
@@ -83,7 +97,8 @@ The default for a change is no new test. Write one only when:
A test that exists must catch a real regression, survive refactoring, run fast,
and be cheap to keep. Never test wiring, constants, defaults, struct shapes,
fakes, or call order. One table-driven test per behaviour; extend the table
fakes, or call order. No test opens a real database or needs Docker; test
SQL semantics at the interceptor or domain level with fakes. One table-driven test per behaviour; extend the table
instead of adding a second test. Prefer making an invalid state unrepresentable
to testing that it is rejected. Test helpers live in `_test.go` files only.
-12
View File
@@ -1,17 +1,5 @@
NOTICE
This product includes software derived from PR-AF
(https://github.com/Agent-Field/pr-af), Copyright AgentField,
licensed under the Apache License, Version 2.0.
Derived portions: internal/domain/workflow/prreview -- the unified diff
parser, change clustering, blast radius, evidence extraction, scoring,
deduplication, line mapping and review-event selection.
PR-AF declares the Apache License, Version 2.0 in its pyproject.toml and
ships no license file. A copy of the License is available at
https://www.apache.org/licenses/LICENSE-2.0
This product includes software from the Cedar project
(https://github.com/cedar-policy/cedar), Copyright Amazon.com, Inc. or its
affiliates, licensed under the Apache License, Version 2.0, as implemented by
+56 -10
View File
@@ -9,6 +9,35 @@ tasks:
cmds:
- task --list
setup:
desc: Install the pinned tooling and dependencies a checkout needs
cmds:
# Report what this task cannot install for you, so a missing Go, Node or
# Docker fails here by name rather than somewhere inside the gate.
- |
set -eu
missing=
for tool in go task node npm docker; do
command -v "$tool" >/dev/null 2>&1 || missing="$missing $tool"
done
if [ -n "$missing" ]; then
echo "warning: not installed (AGENTS.md, Build and gate):$missing" >&2
fi
# The pin is one file, read here and by .github/workflows, so the linter
# running locally and the one running the gate cannot drift apart.
- |
set -eu
want=$(cat {{.ROOT_DIR}}/.golangci-version)
have=$(golangci-lint --version 2>/dev/null | grep -oE '[0-9]+\.[0-9]+\.[0-9]+' | head -1 || true)
if [ "$have" = "${want#v}" ]; then
echo "golangci-lint $want already installed"
else
echo "installing golangci-lint $want (have: ${have:-none})"
GOBIN="$(go env GOPATH)/bin" go install "github.com/golangci/golangci-lint/v2/cmd/golangci-lint@$want"
fi
- task: proto:install
- task: ui:deps
fmt:
desc: Fix and format Go source with the gate's configured formatters
cmds:
@@ -96,11 +125,6 @@ tasks:
cmds:
- golangci-lint run ./...
lint:playbooks:
desc: Validate playbook bindings
cmds:
- go run ./cmd/archied playbooks lint -dir ./playbooks
vuln:
desc: Scan for known vulnerabilities (requires network access)
cmds:
@@ -118,7 +142,7 @@ tasks:
run: once
dir: ui
cmds:
- npm install --no-audit --no-fund
- npm ci --no-audit --no-fund
ui:
desc: Build the dashboard into ui/dist (committed; the Go build embeds it)
@@ -146,6 +170,11 @@ tasks:
cmds:
- bash tools/webhook-send.sh {{.CLI_ARGS}}
bench:prreview:
desc: Run the PR review benchmark on the configured models and write docs/benchmarks/pr-review-*.md (-- -model ALIAS -judge-model REF; --help for flags)
cmds:
- go run ./cmd/archied prbench {{.CLI_ARGS}}
dev:
desc: Run development services and Vite at http://localhost:5173
deps: [dev:daemon, dev:gateway, dev:messaging, dev:state-store, dev:dashboard, dev:ui]
@@ -216,6 +245,22 @@ tasks:
- npm install --no-audit --no-fund
- npm run dev
e2e:up:
desc: Start the local stack without Docker (PostgreSQL 18, services, Vite) and seed it
cmds:
- bash e2e/stack.sh start
- bash e2e/seed.sh
e2e:down:
desc: Stop the local e2e stack
cmds:
- bash e2e/stack.sh stop
e2e:check:
desc: Run the browser checks against the running e2e stack (-- --record VERSION for release evidence)
cmds:
- node e2e/run.cjs {{.CLI_ARGS}}
dev:mobile:
desc: Serve the dashboard on your LAN (token-protected) for phone viewing
deps:
@@ -244,6 +289,7 @@ tasks:
-X github.com/samcharles93/archie-core/internal/buildinfo.Version={{.GATEWAY_VERSION}}
-X github.com/samcharles93/archie-core/internal/buildinfo.Runtime={{.RUNTIME_VERSION}}
sources:
- Taskfile.yml
- "**/*.go"
- go.mod
- go.sum
@@ -256,11 +302,11 @@ tasks:
- "{{.BIN_DIR}}/archie-messaging"
- "{{.BIN_DIR}}/archie-agent"
cmds:
# One flag set for every service, matching scripts/archie-update-install
# and the release workflow, so each binary reports the same version.
# One flag set for every service, matching the release workflow, so each
# binary reports the same version.
- for: [archied, archie-gateway, archie-state-store, archie-ui, archie-messaging]
cmd: go build -ldflags "{{.SERVICE_LDFLAGS}}" -o {{.BIN_DIR}}/{{.ITEM}} ./cmd/{{.ITEM}}
- go build -ldflags "-X github.com/samcharles93/archie-core/internal/app/agentworker.version={{.RUNTIME_VERSION}} -X github.com/samcharles93/archie-core/internal/installtype.buildType=binary" -o {{.BIN_DIR}}/archie-agent ./cmd/archie-agent
cmd: go build -trimpath -ldflags "{{.SERVICE_LDFLAGS}}" -o {{.BIN_DIR}}/{{.ITEM}} ./cmd/{{.ITEM}}
- go build -trimpath -ldflags "-X github.com/samcharles93/archie-core/internal/app/agentworker.version={{.RUNTIME_VERSION}} -X github.com/samcharles93/archie-core/internal/installtype.buildType=binary" -o {{.BIN_DIR}}/archie-agent ./cmd/archie-agent
test:
desc: Run short Go tests with caching
+7 -4
View File
@@ -21,7 +21,6 @@ steps:
- id: plan
type: agent.run
settings:
role: planner
read_only: true
mission: |
Plan this change: {{ task.prompt }}
@@ -135,7 +134,7 @@ _runs without a repository._
| `protect_tests` | bool | Write-block the repository test files. |
| `read_only` | bool | Restrict the agent to read-only tools. |
| `result` | mapping | A JSON Schema object the agent must return; later steps read it as steps.<id>.result. |
| `role` | text | Which configured model role runs the agent, e.g. builder or planner. Empty means builder. |
| `model` | text | The model alias that runs the agent, e.g. default or fast. Empty means default. |
### `command.run`
@@ -247,11 +246,15 @@ _runs without a repository._
| Setting | Type | Meaning |
|---|---|---|
| `inputs` | mapping | |
| `outputs` | mapping | |
| `inputs` | mapping | Callee input names mapped to scalar literals or `inputs.<name>` references to caller inputs. |
| `outputs` | mapping | Callee output names mapped to `outputs.<name>` references to caller outputs; requires `wait: true`. |
| `wait` | bool | Wait for the child run to finish before continuing. |
| `workflow` | text | The workflow to start as a child run. |
The step must declare an `id`: the step's id is the key of the call it starts,
so a retry of that call site returns the child already started instead of
starting a second one.
### `workflow.finish`
_runs without a repository._
+19
View File
@@ -15,10 +15,22 @@ import (
"syscall"
"github.com/samcharles93/archie-core/internal/app/agentworker"
"github.com/samcharles93/archie-core/internal/infrastructure/egress"
)
type workerRunner func(context.Context, agentworker.Settings, *slog.Logger) error
// runHold blocks until signalled and exits 0. It is the keepalive for a Kit
// container that runs no worker: the agent binary is archie's own and is
// mounted into every Kit, so the container needs no binary the image may not
// ship, and holds no environment of its own.
func runHold() int {
ctx, stop := signal.NotifyContext(context.Background(), os.Interrupt, syscall.SIGTERM)
defer stop()
<-ctx.Done()
return 0
}
func main() {
os.Exit(run())
}
@@ -28,6 +40,9 @@ func run() int {
}
func runCommand(args []string, getenv func(string) string, stderr io.Writer, runWorker workerRunner) int {
if len(args) > 0 && args[0] == "hold" {
return runHold()
}
if len(args) > 0 && args[0] == "relay" {
return runRelay(args[1:], stderr, agentworker.RunRelay)
}
@@ -76,6 +91,10 @@ func runCommand(args []string, getenv func(string) string, stderr io.Writer, run
NATSToken: natsToken,
StateStoreTarget: stateStoreURL,
StateStoreToken: stateStoreToken,
StateStoreCA: []byte(getenv("STATE_STORE_CA")),
ModelProxy: getenv(egress.ModelProxyEnv),
ModelProxyHosts: egress.ParseHosts(getenv(egress.ModelProxyHostsEnv)),
ModelProxyCA: getenv(egress.ModelProxyCAEnv),
}, log); err != nil {
return 1
}
-1
View File
@@ -22,7 +22,6 @@ func run() int {
flag.StringVar(&options.Config, "config", configuration.DefaultConfigPath(), "configuration file or directory")
flag.StringVar(&options.Overlay, "config-overlay", "", "configuration overlay file or directory")
flag.StringVar(&options.Listen, "listen", "", "gateway gRPC listen address (defaults to [services.gateway].listen, else 127.0.0.1:8585)")
flag.StringVar(&options.Token, "token", "", "Bearer [REDACTED] required for a non-loopback listener (defaults to [services.gateway].target_token)")
flag.Parse()
showVersion()
ctx, stop := signal.NotifyContext(context.Background(), os.Interrupt, syscall.SIGTERM)
-3
View File
@@ -23,11 +23,8 @@ func run() int {
flag.StringVar(&options.Config, "config", configuration.DefaultConfigPath(), "configuration file or directory")
flag.StringVar(&options.Overlay, "config-overlay", "", "configuration overlay file or directory")
flag.StringVar(&options.Gateway.Target, "gateway-target", "", "archie-gateway gRPC address (defaults to [services.gateway].target, else 127.0.0.1:8585)")
flag.StringVar(&options.Gateway.Token, "gateway-token", "", "bearer token presented to archie-gateway")
flag.StringVar(&options.StateStore.Target, "state-store-target", "", "archie-state-store gRPC address (defaults to [services.state].target)")
flag.StringVar(&options.StateStore.Token, "state-store-token", "", "bearer token presented to archie-state-store")
flag.StringVar(&options.Listen, "listen", "", "MessagingService gRPC listen address (defaults to [services.messaging].listen)")
flag.StringVar(&options.Token, "token", "", "bearer token required from MessagingService callers on a non-loopback listener")
flag.DurationVar(&options.DependencyTimeout, "dependency-timeout", 0, "readiness probe timeout (default 5s)")
flag.DurationVar(&options.ShutdownTimeout, "shutdown-timeout", 0, "shutdown timeout (default 5s)")
flag.Parse()
-1
View File
@@ -34,7 +34,6 @@ func run() int {
flag.StringVar(&options.Config, "config", configuration.DefaultConfigPath(), "configuration file or directory")
flag.StringVar(&options.Overlay, "config-overlay", "", "configuration overlay file or directory")
flag.StringVar(&options.Listen, "listen", "", "state store gRPC listen address (defaults to [services.state].listen, else 127.0.0.1:9090)")
flag.StringVar(&options.Token, "token", "", "bearer token required for a non-loopback listener (defaults to [services.state].target_token)")
flag.StringVar(&options.ReadyAddr, "ready-addr", "", "optional readiness HTTP listen address (e.g. 127.0.0.1:9091)")
flag.Usage = func() {
fmt.Fprintf(flag.CommandLine.Output(), "usage: archie-state-store [flags] # serve the State Store gRPC contract\n\noffline recovery: archie-state-store <backup|restore|validate|rollback> -h | access reset -h\n\n")
+41
View File
@@ -0,0 +1,41 @@
// This file enforces the dashboard's process boundary
// (docs/prds/external-agent-harness.md, "Session contract", Transport): the
// dashboard must not link the container or daemon internals, because only the
// daemon owns containers. Go imports are atomic, so a single reference
// relinks a runtime, and the boundary holds only while something re-measures
// it. archie-core-h3s0 names this check; it was purged with the test suite and
// is restored here.
package main
import (
"context"
"os/exec"
"strings"
"testing"
"time"
)
func TestDashboardDoesNotLinkContainerRuntime(t *testing.T) {
ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
defer cancel()
out, err := exec.CommandContext(ctx, "go", "list", "-deps", ".").Output()
if err != nil {
t.Fatalf("go list -deps .: %v", err)
}
linked := map[string]bool{}
for line := range strings.SplitSeq(strings.TrimSpace(string(out)), "\n") {
linked[line] = true
}
for _, pkg := range []string{
"github.com/samcharles93/archie-core/internal/container",
"github.com/samcharles93/archie-core/internal/app/archied",
"github.com/samcharles93/archie-core/internal/agentexec",
} {
t.Run(pkg, func(t *testing.T) {
if linked[pkg] {
t.Errorf("cmd/archie-ui links %s: only the daemon owns containers", pkg)
}
})
}
}
+1 -4
View File
@@ -27,17 +27,14 @@ func run() int {
flag.StringVar(&options.Config, "config", configuration.DefaultConfigPath(), "configuration file or directory (only [services.*] targets and [web] are read)")
flag.StringVar(&options.Overlay, "config-overlay", "", "configuration overlay file or directory")
flag.StringVar(&options.Listen, "listen", "", "dashboard HTTP listen address (defaults to [web].listen, else 127.0.0.1:8484)")
flag.StringVar(&options.Token, "token", "", "dashboard token; required for a non-loopback listener")
flag.StringVar(&options.TokenFile, "token-file", "", "file holding the dashboard token, minted on first use")
flag.StringVar(&options.OidcIssuer, "oidc-issuer", "", "identity provider issuer URL; requests must then carry a token it signed")
flag.StringVar(&options.OidcAudience, "oidc-audience", "", "audience (resource identifier) this dashboard accepts tokens for; required with -oidc-issuer")
flag.StringVar(&options.OidcClientID, "oidc-client-id", "", "provider client id for browser sign-in; omitted means no sign-in flow")
flag.StringVar(&options.OidcClientSecretEnv, "oidc-client-secret-env", "", "name of the environment variable holding the sign-in client secret")
flag.StringVar(&options.OidcRedirectURL, "oidc-redirect-url", "", "callback URL registered at the provider (path must be /oauth2/callback)")
flag.StringVar(&options.Gateway.Target, "gateway-target", "", "archie-gateway gRPC address (defaults to [services.gateway].target)")
flag.StringVar(&options.Gateway.Token, "gateway-token", "", "bearer token presented to archie-gateway")
flag.StringVar(&options.State.Target, "state-target", "", "archie-state-store gRPC address (defaults to [services.state].target)")
flag.StringVar(&options.State.Token, "state-token", "", "bearer token presented to archie-state-store")
flag.StringVar(&options.Harness.Target, "harness-target", "", "archied harness session gRPC address (defaults to [services.harness].target); empty leaves the setup terminal unwired")
flag.DurationVar(&options.DependencyTimeout, "dependency-timeout", 0, "per-dependency readiness probe timeout (default 5s)")
flag.DurationVar(&options.EventPollInterval, "event-poll-interval", 0, "how often the activity feed polls the State Store for new events (default 1s)")
// BoolFunc rather than BoolVar so an unset flag stays distinguishable
+6 -8
View File
@@ -6,6 +6,7 @@ import (
"github.com/samcharles93/archie-core/internal/app/archied"
"github.com/samcharles93/archie-core/internal/app/archieplaybooks"
"github.com/samcharles93/archie-core/internal/app/prbench"
)
func main() {
@@ -16,18 +17,15 @@ func main() {
if archied.IsSetupArgs(args) {
os.Exit(archied.RunSetup(args[1:], os.Stdin, os.Stdout, os.Stderr))
}
if len(args) > 0 && args[0] == "select-release" {
os.Exit(archied.RunReleaseChannel(args[1:], os.Stdin, os.Stdout, os.Stderr))
}
if len(args) > 0 && args[0] == "update-image" {
os.Exit(archied.RunUpdateImage(args[1:], os.Stderr))
}
if len(args) > 0 && args[0] == "update-topology" {
os.Exit(archied.RunUpdateTopology(args[1:], os.Stdout, os.Stderr))
if len(args) > 0 && args[0] == "update" {
os.Exit(archied.RunUpdate(args[1:], os.Stdout, os.Stderr))
}
if len(args) > 0 && args[0] == "status" {
os.Exit(archied.RunStatus(args[1:], os.Stdout, os.Stderr))
}
if len(args) > 0 && args[0] == "prbench" {
os.Exit(prbench.Main(args[1:], os.Stdout, os.Stderr))
}
if len(args) > 0 && args[0] == "playbooks" {
os.Exit(archieplaybooks.Run(args[1:], os.Stderr))
}
+29 -26
View File
@@ -17,9 +17,6 @@
# database_url = "postgres://archie:archie@127.0.0.1:5432/archie?sslmode=disable"
# skills_dir = "~/.local/share/archie/skills" # SKILL.md search path; defaults under work_dir when unset
# workflow_routing_file = "~/.local/share/archie/workflow-routing.yaml" # kind->workflow rebinding (bug/feature)
# workflow_labels_file = "~/.local/share/archie/workflow-labels.yaml" # arbitrary forge label -> workflow (EDA playbook slice 2)
# playbook_dirs = ["~/.local/share/archie/workflows"] # list of dirs of *.yaml binding files, merged with the two above (slice 3); a key bound twice across sources is a load failure
poll_interval = "60s"
label = "archie"
@@ -63,7 +60,8 @@ token = { engine = "env", key = "ARCHIE_GITHUB_TOKEN" }
# intake per identity is a pending migration, not a setting.
# [[identities]] runs more than one bot identity from a single daemon, each
# with its own forge/dispatch/models/providers/budgets/notify. Intake
# with its own forge account and repository membership. Runtime settings
# are shared and edited through the control plane. Intake
# settings are rejected per identity, and [forge].intake = "webhook"/"both"
# is rejected alongside [[identities]]. Minimal shape:
# [[identities]]
@@ -146,16 +144,9 @@ class = "ollama"
# "openrouter/openai/gpt-5.6-sol",
# ]
# Optional self-hosted update adapter. Both values are argv arrays (never
# shell snippets). The check command must write a releaseupdate.Snapshot JSON
# document to stdout; configure the installer as well to show Approve/Defer.
# [chat.telegram]
# The bot token from @BotFather, resolved through a configured secret engine.
# token = { engine = "bws", key = "TELEGRAM_BOT_TOKEN" }
# Run by archie-messaging, which serves /update, so these commands execute on
# whichever host runs that process.
# update_check_command = ["/usr/local/bin/archie-update-check"]
# update_install_command = ["/usr/local/bin/archie-update-install"]
# Email and the generic inbound webhook are archipelago channel extensions
# (channels/email, channels/webhook): install, accept and enable them from the
@@ -255,12 +246,14 @@ listen = "127.0.0.1:8484"
# [memory]
# engine = "builtin" # "builtin" is the only engine implemented so far
# Notifications for human-in-the-loop moments
# (approval requests, gate failures, parked tasks).
# Archie POSTs a JSON payload to the webhook
# wire it to n8n, email, Slack, or any HTTP endpoint.
# Operator alerts (parked tasks, failed runs, unhealthy services, missing
# credentials) are listed in the dashboard. Archie also POSTs each raised and
# resolved alert as JSON to the webhook: wire it to n8n, email, Slack, or any
# HTTP endpoint. Failed deliveries are retried and recorded on the alert.
[notify]
# webhook = "https://n8n.example.com/webhook/archie"
# min_severity = "warning" # info, warning or action: the lowest severity delivered
# secret = {engine = "env", key = "ARCHIE_NOTIFY_SECRET"} # signs X-Archie-Signature (HMAC-SHA256)
[budgets]
max_steps = 60
@@ -268,6 +261,18 @@ wall_clock = "30m" # one agent call
gate_max_failures = 5
task_wall_clock = "4h" # a whole task run; keep containers.max_uptime at or above it
# TLS: every gRPC hop between services uses certificates from your own CA
# (step-ca, cert-manager, Vault, openssl). Each process loads
# cert_dir/<service>.pem and cert_dir/<service>-key.pem, whose DNS SAN is its
# service name: archied, archie-gateway, archie-messaging, archie-state-store,
# archie-ui. Every peer must chain to ca_file. Certificates are reloaded when
# they change. Without [tls], services may only listen on and dial loopback
# addresses, task containers cannot reach the State Store, and the harness
# setup terminal is off.
# [tls]
# ca_file = "/etc/archie/tls/ca.pem"
# cert_dir = "/etc/archie/tls"
# Services selects contract adapters at composition time.
[services.gateway]
target = "127.0.0.1:8585" # archie-gateway's gRPC address (always remote once set)
@@ -276,11 +281,6 @@ target = "127.0.0.1:8585" # archie-gateway's gRPC address (always remot
# # to the target above, so a default deployment's client and
# # server cannot drift onto different ports. Set it when the
# # host already owns that port.
# target_token = "..." # Bearer [REDACTED] for a NON-loopback listener. A loopback listener
# # (127.0.0.1/::1/localhost) needs no token. A non-loopback
# # listener without a token fails closed, as does a
# # non-loopback dial without one. The same token is
# # presented as GATEWAY_TOKEN by remote consumers.
[services.state]
target = "127.0.0.1:9090" # REQUIRED. The State Store gRPC address archied and
@@ -292,12 +292,15 @@ target = "127.0.0.1:9090" # REQUIRED. The State Store gRPC address arch
# # above, so a default deployment's client and server cannot
# # drift onto different ports. Move it when a port is already
# # taken on the host.
# target_token = "..." # bearer token for a NON-loopback target. A loopback target
# # (127.0.0.1/::1/localhost) needs no token. A non-loopback
# # target without a token fails closed. The same token is injected as STATE_STORE_TOKEN into
# # archie-agent containers so the agent authenticates to the
# # same remote store. See deployments/README.md for how to
# # start archie-state-store.
# The daemon's harness session contract: the dashboard's setup terminal opens
# a PTY into an ephemeral Kit container over it. Both ends are optional and
# the dashboard's setup terminal degrades to 503 without them: the daemon
# serves the contract only when [tls] is configured, and archie-ui dials it
# only when a target is set.
# [services.harness]
# target = "127.0.0.1:8587" # The address archie-ui dials. Empty leaves the
# # setup terminal unwired.
# NATS is the sole autonomous workflow handoff. mode changes only where the
# broker runs; every workflow still runs in a managed task container:
+36 -36
View File
@@ -55,14 +55,9 @@ as a host binary. Pull it from GHCR: `ghcr.io/samcharles93/archie-agent`.
./archied -config ~/.config/archie/config.toml
```
Dashboard reachability comes from `[web].listen` (default
`127.0.0.1:8484`, which needs no dashboard token). For a non-loopback
bind pass `-listen` with `-token` or `-token-file` to mint and persist
one. `archie-ui` reads `GATEWAY_TOKEN` / `STATE_STORE_TOKEN` from its
process environment only — it has no secret registry, so when those
tokens live in a secret manager, whatever starts `archie-ui` must export
them into its environment (the `[services.*].target_token` values in
`config.toml` are the other supported source).
Open `[web].listen` (default `127.0.0.1:8484`) to configure the organisation,
workspace and first administrator account. Remote access requires an HTTPS
reverse proxy. Complete setup before exposing the dashboard publicly.
4. For a persistent, 24/7 deployment instead of a foreground run, follow the
systemd user-unit runbook:
@@ -90,42 +85,47 @@ an install (or an upgrade across v1.30.0) needs its unit as well -- see
### Native self-update
`archie-update-install` downloads the approved release's Linux/amd64 archive
and verifies its `SHA256SUMS` entry before executing or installing it. Releases
must include `release.json`, all host binaries, and the three updater scripts.
Older releases without this contract cannot be installed through the artifact
path. `--from-source` (or `ARCHIE_UPDATE_SOURCE_BUILD=1`) explicitly clones and
builds the selected release instead.
Install on Settings → Status runs `archied update install -version X`, which
downloads the release's Linux/amd64 zip and verifies it against `SHA256SUMS`
before unpacking it. Every host binary in the zip must report `X`. It then pulls
the agent image `release.json` pins by digest, backs up the database to
`db.pre-update.dump` in the state directory, stages the new binaries beside the
live ones and hands the transaction to `archied update watchdog` in a process of
its own (a transient systemd user unit). Nothing that runs changes until the
watchdog takes over.
The update check command accepts `--channel stable` (default), `--channel next`
(includes prereleases), or `--channel exact-pin --pin 1.2.3`. Set these arguments
in the control plane's channel settings alongside the existing check command.
`archied update check` finds the newest GitHub release with a linux-amd64 zip on
the channel chosen under Settings → Status → Updates: stable, next (includes
prereleases) or a pinned version. From a shell it reads
`ARCHIE_UPDATE_RELEASE_CHANNEL` and `ARCHIE_UPDATE_PIN`, or `-channel` and `-pin`.
The notification destination remains separate from the release channel.
For scheduled, unattended runs, `archie-update-install --auto` discovers the
selected release and refuses if the daemon reports work in flight. It checks
again immediately before promotion. A scheduler can set
`ARCHIE_UPDATE_RELEASE_CHANNEL` and `ARCHIE_UPDATE_PIN`; nothing schedules updates
by default. All services stop through their ordinary graceful shutdown path.
`release.json` supplies the service order and required config sections to both
first-install and self-update. A missing unit or config section produces an
ordered migration plan and refuses installation. Apply that plan before retrying.
Automatic topology edits are intentionally unsupported.
The watchdog switches staged executables by rename, restarts services, and
requires the candidate daemon's existing readiness registry to remain healthy
for 120 seconds. Agent images come from the release manifest by digest. The
image switch updates the control-plane policy, preserving other settings;
rollback restores the prior image policy with conflict detection. Source builds
use the built image's immutable image ID.
The watchdog stops the units in reverse `release.json` order, switches the
staged executables by rename, switches the control-plane image policy (only the
image field, with conflict detection), and restarts the units in order through
`systemctl --user`. archied must report the new version and stay healthy for
120 seconds. Otherwise the watchdog restores the previous binaries and image
policy. Data is not rolled back unless the new release migrated the database:
then the previous binaries cannot serve it, so the pre-update backup is restored
and the outcome report says that every write since the update began is gone.
Without a systemd user manager the watchdog switches binaries and image policy
but restarts nothing, and the install reports that the services need a manual
restart.
Transactions live in `${XDG_STATE_HOME:-~/.local/state}/archie/update`, outside
the binary directory. `phase`, `environment`, `image.json` (when applicable),
and `previous/` describe the transaction. A lock reports the owning process.
On daemon boot, an interrupted transaction schedules independent recovery before
normal startup. To recover manually, run `transaction/watchdog --recover` from
that state directory. A failed recovery keeps the journal for another attempt.
the binary directory: `transaction/journal.json`, `transaction/previous/`, and
`transaction/archied`, the frozen copy of the archied that wrote the journal. On
daemon boot, an interrupted transaction is
handed to that frozen watchdog with `-recover` before normal startup: a
transaction that never started is discarded, one that was switching or observing
is rolled back, and a decided one is finished. To recover manually, run
`transaction/archied update watchdog -recover -bin-dir ~/.local/bin` from that
state directory. A failed recovery keeps the journal for another attempt.
Database snapshots remain available for manual recovery. Schema migrations and
PostgreSQL data are never automatically reversed.
The database backup stays in the state directory after a committed update for
manual recovery with `archie-state-store restore`.
+18 -13
View File
@@ -52,10 +52,10 @@ PostgreSQL database `database_url` names and serves the `StateStore` gRPC
contract. BOTH `archied` and `archie-gateway` dial it via
`[services.state].target` and never open the task tables themselves. The templates set
`[services.state]` accordingly: a host-only agent reaches a loopback bind
(`127.0.0.1:9090`) with no token, while a container-mode agent needs the
Docker bridge gateway address plus a matching bearer token (`target_token` /
`--token` / `STATE_STORE_TOKEN`), because a container cannot reach the host's
loopback.
(`127.0.0.1:9090`), while a container-mode agent needs the Docker bridge
gateway address, because a container cannot reach the host's loopback. A
non-loopback address needs `[tls]` certificates from your own CA; see
`config.example.toml`.
## Running the dashboard
@@ -69,24 +69,29 @@ default). Start it after the Gateway and the State Store:
archie-ui -config ~/.config/archie/config.toml
```
A non-loopback `[web].listen` requires a dashboard token (`-token`, or
`-token-file` to mint and persist one); a loopback bind needs none.
Open the dashboard to configure the organisation, workspace and first administrator account.
Use HTTPS through a reverse proxy for remote access, and complete setup before exposing it publicly.
`archied` runs no dashboard listener at all, so webhook capture
(`POST /webhooks/capture/{source}`) is received by archie-ui and the
binding-dispatch loop in `archied` consumes what arrives from the shared
State Store.
Auth note: unlike `archied`, archie-ui has no secret registry --
`GATEWAY_TOKEN`/`STATE_STORE_TOKEN` resolve from its process environment
only (or from `[services.*].target_token` in `config.toml`), so whatever
starts archie-ui must export those tokens into its environment. See the UI
Service unit in `systemd-user-service.md` for the worked example.
The dashboard speaks plain HTTP. A non-loopback bind therefore carries the
session cookie and the provider token in clear over the network, and
archie-ui warns about exactly that at startup. Terminate TLS in front of it
(a reverse proxy or the tunnel itself) and set
`[web].trust_forwarded_headers = true` to declare the proxy and silence the
warning; leaving it off while exposing the port directly is the case the
warning names.
Auth note: archie-ui authenticates to its dependencies with its `[tls]`
certificate (`archie-ui.pem`).
```bash
# loopback (host-only agent)
archie-state-store -config ~/.config/archie/config.toml -listen 127.0.0.1:9090
# or container-mode: bind the bridge gateway with a token
archie-state-store -config ~/.config/archie/config.toml -listen 172.17.0.1:9090 -token <token>
# or container-mode: bind the bridge gateway ([tls] required)
archie-state-store -config ~/.config/archie/config.toml -listen 172.17.0.1:9090
```
## Verifying the daemon
+7 -5
View File
@@ -26,13 +26,15 @@ target = "127.0.0.1:8585"
# The State Store gRPC service owns the PostgreSQL task data; archied and
# archie-gateway dial it and never open those tables themselves.
# This deployment runs the agent in containers, so target is the Docker bridge
# gateway (172.17.0.1) the container can reach, with a matching target_token
# that archie-state-store validates. Host-only agent: use 127.0.0.1:9090 with
# no token.
# This deployment runs the agent in containers, so the target is the Docker
# bridge gateway (172.17.0.1) the container can reach. A non-loopback address
# needs [tls]; see config.example.toml.
[services.state]
target = "172.17.0.1:9090"
target_token = "change-me-state-store-token"
[tls]
ca_file = "/etc/archie/tls/ca.pem"
cert_dir = "/etc/archie/tls"
[forge]
type = "gitea"
+3 -5
View File
@@ -18,13 +18,11 @@ target = "127.0.0.1:8585"
# The State Store gRPC service owns the PostgreSQL task data; archied and
# archie-gateway dial it and never open those tables themselves.
# This deployment runs the agent in containers, so target is the Docker bridge
# gateway (172.17.0.1) the container can reach, with a matching target_token
# that archie-state-store validates. Host-only agent: use 127.0.0.1:9090 with
# no token.
# This deployment runs the agent in containers, so the target is the Docker
# bridge gateway (172.17.0.1) the container can reach. A non-loopback address
# needs [tls]; see config.example.toml.
[services.state]
target = "172.17.0.1:9090"
target_token = "change-me-state-store-token"
[models]
triage = "openai/gpt-6-luna"
+3 -8
View File
@@ -25,16 +25,11 @@ target = "127.0.0.1:8585"
# The State Store gRPC service owns the PostgreSQL task data. archied and
# archie-gateway dial it here and never open those tables themselves.
# This deployment runs the
# agent in containers, so the target must be the Docker bridge gateway address
# (172.17.0.1) that the container can reach, and a non-loopback target requires
# a matching target_token that archie-state-store validates -- start it with the
# same token (e.g. `archie-state-store ... -listen 172.17.0.1:9090 -token
# change-me-state-store-token`). For a host-only agent (no [containers]) use the
# loopback target 127.0.0.1:9090 with no token.
# This deployment runs the agent in containers, so the target is the Docker
# bridge gateway (172.17.0.1) the container can reach. A non-loopback address
# needs [tls]; see config.example.toml.
[services.state]
target = "172.17.0.1:9090"
target_token = "change-me-state-store-token"
[forge]
type = "github"
+13 -36
View File
@@ -20,10 +20,9 @@ The State Store unit is `archie-state-store.service`. It owns the task data in
PostgreSQL and serves the `StateStore` gRPC contract. BOTH `archied` and
`archie-gateway` dial it via `[services.state].target` (see the `archied`
config below) and never open the task tables themselves. A
host-only agent uses the loopback bind with no token below; a container-mode
agent needs the Docker bridge gateway address plus a bearer token via
`--token`/`STATE_STORE_TOKEN` or `[services.state].target_token`, and a
non-loopback bind fails closed without one:
host-only agent uses the loopback bind below; a container-mode agent needs the
Docker bridge gateway address, and a non-loopback bind needs `[tls]` (see
`config.example.toml`):
`install.sh` writes `archie-state-store.service` with that loopback bind.
@@ -36,14 +35,12 @@ by `archied`, since the v1.30.0 extraction. `archied` starts no gateway for them
and logs nothing when one is absent, so a host that updates across v1.30.0
without this unit runs on with dead channels and a daemon that looks perfectly
healthy. `install.sh` writes this unit on a fresh install;
`scripts/archie-update-install` refuses an
`archied update install` refuses an
update when a host is missing a unit and names what is missing, which now fires
for a host that drifted or predates this release rather than the normal path.
It dials the Gateway for the chat contract and the State Store for the stored
channel settings, and reads the same `config.toml`. Like `archie-ui` it resolves
`GATEWAY_TOKEN` and `STATE_STORE_TOKEN` from its process environment, hence the
`EnvironmentFile`:
channel settings, and reads the same `config.toml`, including `[tls]`.
`install.sh` writes `archie-messaging.service`. Its `After=` and `Wants=`
lines only tidy the start sequence: nothing depends on that order, because
@@ -53,46 +50,26 @@ each service waits for the peers it needs.
The dashboard is served by the standalone UI Service, not by `archied`. It reads the same `config.toml`, dials
the Gateway and State Store targets already in it, and binds `[web].listen`.
A non-loopback bind requires a dashboard token (`-token`, or `-token-file` to
mint and persist one);
Open the dashboard to configure the organisation, workspace and first administrator account.
Complete setup before exposing it publicly; remote access requires an HTTPS reverse proxy.
**Auth note:** unlike `archied`,
`archie-ui` has no secret registry — it resolves `GATEWAY_TOKEN` and
`STATE_STORE_TOKEN` from its **process environment only** (or from the
`target_token` values in `config.toml`). Whatever starts `archie-ui` is
therefore responsible for exporting those tokens into its environment: keep
them in `~/.config/archie/env` (loaded by the `EnvironmentFile` line below),
not only in a secret manager. A deployment that authenticates `archied`
through bws/Vault but exports nothing for `archie-ui` leaves the dashboard
unable to dial its dependencies.
**Auth note:** `archie-ui` authenticates to the Gateway, State Store and
harness with its `[tls]` certificate (`archie-ui.pem`) from `config.toml`.
For supervision, `GET /healthz` on `[web].listen` is the process's liveness
probe (token-free by design); the authenticated readiness surface lives at
the dashboard's health endpoint behind the token. A watchdog can curl
the dashboard's health endpoint behind account authentication. A watchdog can curl
`/healthz` without credentials.
`install.sh` writes `archie-ui.service`. It has no secret registry of its own,
so its token handling is the `EnvironmentFile` note above.
Concretely, for a deployment whose dependency tokens live outside
`config.toml`, `~/.config/archie/env` carries the two names and the unit
loads it; the dashboard token itself comes from `-token-file` (there is no
environment variable for it):
```bash
# ~/.config/archie/env
GATEWAY_TOKEN=... # presented to archie-gateway
STATE_STORE_TOKEN=... # presented to archie-state-store
```
`install.sh` writes `archie-ui.service`. Browser accounts are configured through the dashboard.
`After=` orders the start, it does not wait for the store to bind. The UI
process closes that gap itself: its event pump retries priming with backoff
until the State Store answers, so the activity feed goes live once the store
is up rather than staying history-only until the next restart.
For a non-loopback bind, override the installed unit's `ExecStart` with
`-token-file %h/.local/share/archie/web-token` (or `-token`) in a drop-in rather
than editing the installer's copy in place.
For remote access, configure a TLS reverse proxy and the instance public URL in
Access → Authentication. Enrol passkeys from Your account on HTTPS or localhost.
---
+10 -2
View File
@@ -44,10 +44,18 @@ services:
# ── NATS JetStream ────────────────────────────────────────────────
nats:
image: nats:2.14-alpine3.22
command: -js -m 8222
# The client port is published on every interface (agent containers reach
# it over the bridge), so it refuses to start without a token. Set
# NATS_TOKEN here and name the same value in [nats] token_env.
entrypoint:
- sh
- -c
- '[ -n "$$NATS_TOKEN" ] || { echo "NATS_TOKEN is required" >&2; exit 1; }; exec nats-server -js -m 8222 --auth "$$NATS_TOKEN"'
environment:
NATS_TOKEN: ${NATS_TOKEN:-}
ports:
- "4222:4222" # client
- "8222:8222" # monitoring (http://localhost:8222)
- "127.0.0.1:8222:8222" # monitoring (http://localhost:8222)
volumes:
- nats_data:/data
restart: unless-stopped
+24
View File
@@ -0,0 +1,24 @@
# PR review benchmark
```
task bench:prreview
```
Reviews the 38 runnable Martian pull requests once each at `deep` depth with
the production pipeline, judges the posted comments against the golden
comments in `internal/app/prbench/martian.json`, and writes
`docs/benchmarks/pr-review-<timestamp>.md` (recall per PR, overall recall,
precision, models, archie version). Commit that file with the run.
- Models come from the config at `~/.config/archie/config.toml`: the reviewer
and judge both default to its `default` alias. Override with
`task bench:prreview -- -model ALIAS_OR_PROVIDER/MODEL -judge-model ALIAS_OR_PROVIDER/MODEL`.
Pick a judge other than the reviewer's model for an independent judgement.
Aliases set only in the dashboard are not in the file; pass a
`provider/model` ref instead.
- Pull requests are public, so no credential is needed. `GH_TOKEN` or
`GITHUB_TOKEN`, if set, only raises the GitHub API rate limit.
- `-problems ID,ID` runs a partial smoke run; its report says it is not the
acceptance run. A full run reports PASS when nothing failed and golden
micro-recall is at least 0.70.
- It is not part of `task test` and posts nothing to any pull request.
+30
View File
@@ -64,7 +64,22 @@
"/news/archie/1.47.0/": "/news/1.47.0/",
"/news/archie/1.47.1/": "/news/1.47.1/",
"/news/archie/1.47.2/": "/news/1.47.2/",
"/news/archie/1.47.3/": "/news/1.47.3/",
"/news/archie/1.48.0/": "/news/1.48.0/",
"/news/archie/1.48.1/": "/news/1.48.1/",
"/news/archie/1.49.0/": "/news/1.49.0/",
"/news/archie/1.49.1/": "/news/1.49.1/",
"/news/archie/1.49.2/": "/news/1.49.2/",
"/news/archie/1.49.3/": "/news/1.49.3/",
"/news/archie/1.49.4/": "/news/1.49.4/",
"/news/archie/1.5.0/": "/news/1.5.0/",
"/news/archie/1.50.0/": "/news/1.50.0/",
"/news/archie/1.50.1/": "/news/1.50.1/",
"/news/archie/1.51.0/": "/news/1.51.0/",
"/news/archie/1.51.1/": "/news/1.51.1/",
"/news/archie/1.52.0/": "/news/1.52.0/",
"/news/archie/1.53.0/": "/news/1.53.0/",
"/news/archie/1.54.0/": "/news/1.54.0/",
"/news/archie/1.6.0/": "/news/1.6.0/",
"/news/archie/1.7.0/": "/news/1.7.0/",
"/news/archie/1.8.0/": "/news/1.8.0/",
@@ -145,7 +160,22 @@
"/news/archied/1.47.0/": "/news/1.47.0/",
"/news/archied/1.47.1/": "/news/1.47.1/",
"/news/archied/1.47.2/": "/news/1.47.2/",
"/news/archied/1.47.3/": "/news/1.47.3/",
"/news/archied/1.48.0/": "/news/1.48.0/",
"/news/archied/1.48.1/": "/news/1.48.1/",
"/news/archied/1.49.0/": "/news/1.49.0/",
"/news/archied/1.49.1/": "/news/1.49.1/",
"/news/archied/1.49.2/": "/news/1.49.2/",
"/news/archied/1.49.3/": "/news/1.49.3/",
"/news/archied/1.49.4/": "/news/1.49.4/",
"/news/archied/1.5.0/": "/news/1.5.0/",
"/news/archied/1.50.0/": "/news/1.50.0/",
"/news/archied/1.50.1/": "/news/1.50.1/",
"/news/archied/1.51.0/": "/news/1.51.0/",
"/news/archied/1.51.1/": "/news/1.51.1/",
"/news/archied/1.52.0/": "/news/1.52.0/",
"/news/archied/1.53.0/": "/news/1.53.0/",
"/news/archied/1.54.0/": "/news/1.54.0/",
"/news/archied/1.6.0/": "/news/1.6.0/",
"/news/archied/1.7.0/": "/news/1.7.0/",
"/news/archied/1.8.0/": "/news/1.8.0/",
+75
View File
@@ -1,4 +1,79 @@
[
{
"version": "1.54.0",
"date": "2026-10-16",
"body": "**Upgrade note:** services now authenticate each other with TLS certificates from your own CA instead of shared tokens. Before upgrading a host whose services listen on anything but loopback (any host running agent containers), issue a certificate per service (`archied`, `archie-gateway`, `archie-messaging`, `archie-state-store`, `archie-ui`, each with its name as the DNS SAN and both server and client auth), add `[tls] ca_file` and `cert_dir` to `config.toml`, and remove every `target_token`, the `*_TOKEN` service variables and the `--token`/`--*-token` flags, including archie-ui's `-token` and `-token-file`. A unit that still passes a removed flag fails to start. See `config.example.toml`.\n\n### archied\n\n- Services authenticate each other with mutual TLS. Each process presents `<cert_dir>/<service>.pem`, peers must chain to `ca_file`, and a caller is identified by its certificate rather than by what it claims. Renewed certificates are picked up without a restart. Without `[tls]`, services listen on and dial loopback only. The harness setup terminal requires `[tls]`.\n- Every gRPC service caps the streams a connection holds, times out stalled handshakes, releases a dead client's streams, and drains for at most 10 seconds on shutdown. Messaging exits when its listener fails instead of staying up without delivering.\n- The dashboard has user accounts. First run opens browser setup for the organisation, workspace and first administrator; people then sign in with a password, a passkey or your identity provider, and account administration lives under Access. The shared dashboard token is gone; reach the dashboard remotely through an HTTPS reverse proxy.\n- `archied update install` updates a host in one transaction: it verifies the release, stages every binary, pulls the agent image by digest and backs up the database, then restarts the services in order and rolls everything back if archied is not healthy for two minutes.\n- Each workflow runs as the identity it declares, checked on every attempt along with the identity that started it. A workflow that declares none is refused, a called workflow runs as its caller's declared identity, and shipped workflows run as the instance root.\n- Test-run a workflow draft before saving it, with forge and push writes suppressed and recorded per step. The History tab lists a workflow's revisions, diffs them as YAML and on the canvas, restores one into the editor, and shows which revision each run used.\n- Start a task from the dashboard. The dashboard home shows activity and setup, and a running task has a read-only Sandbox tab that follows its container.\n- The topbar's Jump to is a command palette, settings lists open as accordions that wait for a new item to be saved, and history restores land as a draft applied from the save bar.\n- Labels, hints, alerts and empty states say what is wrong and what to do, without tutorial or marketing prose.\n- Org isolation is tighter: harness secrets, identity writes, principals, playbook dispatch and model selections in chat stay in the caller's org, instance settings are written by the system org only, and creating a source or reloading a channel is authorized against the target.\n- Binding dispatch claims its capture and enqueues the task atomically, so a capture cannot start two runs, and the run-time workflow check sees the binding event's signature and address.\n- Service-unhealthy alerts count their grace from the last report and also cover a service that went silent, and a restart no longer clears an alert still inside its grace. An invalid stored access policy raises an alert until it validates.\n- A failed cancel of a called workflow no longer leaves the caller running, retention spares captures still waiting to be dispatched, and EDA playbooks load and report at start instead of on first dispatch.\n- Chat model selection works again for provider models.\n\n### archie-agent\n\n- Each run acts as its workflow's declared identity.\n- A task snapshot carries only the model limits its aliases resolve to.\n- A build step's \"no changes\" result comes from its writable stages and clears once a commit lands.\n- Test runs suppress forge and push writes and record what they would have done.\n- The default persona is a reporting-focused identity prompt, without the one-question limit in chat."
},
{
"version": "1.53.0",
"date": "2026-10-11",
"body": "### archied\n\n- Every organisation's data is its own. Chats, tasks, events, transitions, steps, tool calls, run statistics, event sources, event types, mappings, bindings and agent memory are read and written in the org that owns them, and a request can no longer write into another org. Rows written before orgs existed move to their owner's org on upgrade.\n- An instance admin decides what each org may reach: the models it can use, and whether it may register providers under its own credentials. Org Settings gains a Models tab and instance Models gains Org access. The system org is never restricted; another org uses nothing until it is granted.\n- Models are addressed by alias. Aliases replace model roles, one resolver serves every purpose (agent steps, implement, PR review, MCP sampling, chat, embedding, transcription), and an unknown alias fails instead of falling back to a default. The step editor and the chat picker offer aliases, and a workflow naming one that does not exist is refused when saved.\n- Chat sessions belong to one org; a chat resolves its models in the caller's org and is billed to it, and a task run from chat posts its outcome back to that chat.\n- Add a member by provider subject, or by email as a placeholder that the first sign-in the provider vouches for claims.\n- The workflow canvas authors parallel branches as lanes, adds a switch step that runs the case matching a reference, and edits a call's interface. A step shows its role, its number and its run trace.\n- The workflows list shows each workflow's running count and latest run result, the canvas reveals its data-edge handles on hover, and restoring a version asks for confirmation.\n- Store packages can auto-update: a package given the policy is moved to the catalogue's newest digest by a State Store ticker. An update asking for more authority than was accepted waits as a pending update with the running version untouched; approving it accepts the new authority, and a rollback restores the previous version with the authority accepted against it.\n- Model provider keys never enter an agent container: the runtime reaches its providers through the egress proxy, which holds the keys. A task's own credential can no longer replace the workflow definition its run is pinned to, and a package workflow's triggers, tools and forge calls are bounded by the authority accepted for the package.\n- The dashboard warns when it is reachable from off-host over plain HTTP, and the wildcard listen address counts as off-host rather than loopback. Instance health, logs, config and identities, the setup checklist and the skill catalogue are system-org only, and API requests are refused until the access chain has loaded.\n- `/approve` reaches Telegram; the router served it but nothing published it.\n- A worktree a killed container left root-owned is taken back by the daemon, and a PR review's benchmark runner works again on configured models and public PRs.\n- An unserved `/api` path returns 404 instead of the dashboard page, and an input group's select no longer loses clicks to its addon.\n\n### archie-agent\n\n- The runtime reaches model providers through the egress proxy, so provider keys stay out of the container.\n- Agent steps resolve model aliases and their own org's models and credentials, failing on an unknown alias instead of falling back.\n- A switch step runs the case matching its reference.\n- A call waiting on another step is reconciled from durable steps, so it resumes after a daemon restart rather than stranding."
},
{
"version": "1.52.0",
"date": "2026-10-09",
"body": "### archied\n\n- The harness page's setup terminal opens a real session: it starts an ephemeral container from a Kit profile, runs the CLI's own login and captures the OAuth tokens at the provider's token endpoint. Point the dashboard at the daemon with `[services.harness].target` and a `HARNESS_TOKEN` and it stops answering 503; the terminal is created at your browser's size. Takeover of a live task is not in this release.\n- A credential's token endpoint is refused when the run is not granted that service, instead of proxy-ing the provider's real tokens into the container.\n- Settings → Profiles, Credential bindings and Playbooks are editable from the dashboard, and curator definitions can be changed and applied without a restart.\n- Settings say whether each one applies live or needs a restart, show what is consuming resources, and report whether an apply landed.\n- Status shows each service's dependency checks, the agent workers and the broker.\n- The Logs page reads recent logs from every service, not just the daemon.\n- Captures and webhook sources are scoped to the org that owns them: the list, retention and delivery authorization all act in the right org, and a workflow definition resolves in its own org.\n- Chat settings are attributed to the signed-in user, and Settings → Tokens explains when a personal token cannot be created.\n- A channel enabled after Archie started works immediately; a channel that fails on its own recovers; a failed channel is no longer reported as enabled, and stored channel settings no longer save duplicate or unnormalized values.\n- The task table sorts, the command palette no longer scrolls sideways or loses its highlight, unknown routes render a not-found page, and configuration reload failures are shown instead of passing silently.\n- A scheduled job whose payload names a dangerous command is refused.\n- A disabled model provider is never served from the ambient API key.\n- `generate_video` is backed by ai-sdk's MiniMax video provider.\n- A dashboard token is no longer written to the startup log.\n\n### archie-agent\n\n- `workflow.call`: a call is keyed on its declared step path, its key is freed once the callee reaches a terminal state, and a caller waiting on a callee releases its capacity instead of holding it.\n- `command.run` honours step retry and `on_failure`.\n- A value passed between steps keeps its declared type, and optional result booleans are enforced only when the workflow declares them.\n- Capture tools have their results checked against the declared schema.\n- A single-cycle run waits for its in-flight tasks before exiting, and identity tasks inherit live settings.\n- Kit containers run Docker's init as PID 1, so orphaned processes are reaped instead of accumulating inside a long task."
},
{
"version": "1.51.1",
"date": "2026-10-08",
"body": "### archied\n\n- Chat with OpenAI's gpt-6 models (gpt-6-luna and the rest) works with tools; a turn failed with \"Function tools with reasoning_effort are not supported\"."
},
{
"version": "1.51.0",
"date": "2026-10-08",
"body": "### archied\n\n- A secret engine enabled after Archie started (Bitwarden, for example) now supplies its keys straight away; providers that depended on it no longer stay disabled until a restart.\n- The dashboard chat has a microphone button: record a voice message and it is transcribed like a Telegram voice note. It shows only when a `transcription` model role is set under Settings → Models.\n- A file Archie sends in the dashboard chat is a download link for an hour, instead of a note that it can't be delivered.\n- Typing `/model` or `/personality` on its own in the dashboard chat offers the choices as buttons."
},
{
"version": "1.50.1",
"date": "2026-10-08",
"body": "### archied\n\n- When an update is rolled back, the log names the health check that failed instead of only \"unhealthy\"."
},
{
"version": "1.50.0",
"date": "2026-10-08",
"body": "### archied\n\n- Tools that need your approval ask for it again, and Archie can ask you a question or offer choices mid-reply. Since the services were split these never reached Telegram or the dashboard: approval-gated tools failed and questions were never asked.\n- The dashboard chat shows approvals as Approve / Always this session / Deny, choices as buttons, and questions with a reply field.\n- `/update` in the dashboard chat lists what's available with Install and Not now.\n- Settings → Status → Versions installs the update it shows; its Install button sent an empty request."
},
{
"version": "1.49.4",
"date": "2026-10-07",
"body": "### archied\n\n- Settings → Status and Settings → Extensions are split into tabs instead of one long page; the Versions table's status column ends at the card's edge.\n- Tools & MCP has one \"Add MCP server\" button; the transport is chosen in the form, and an unnamed new server is not an unsaved change.\n- Settings → Repositories shows the add field when no repositories are configured, instead of a blank page.\n- The unsaved-changes bar always rests at the bottom of the window."
},
{
"version": "1.49.3",
"date": "2026-10-07",
"body": "### archied\n\n- Settings → Status → Versions shows the Gateway as \"Update available\" with an Install button when a newer release exists; it read \"Unknown\" with no way to install. The Runtime row stays \"Unknown\" until an agent task has run.\n- The release channel buttons sit at the Updates card's right edge instead of mid-row."
},
{
"version": "1.49.2",
"date": "2026-10-07",
"body": "### archied\n\n- Carries the 1.49.0 and 1.49.1 changes, whose builds were never published. The update check reads GitHub releases directly and still offers only releases with a verifiable linux-amd64 zip."
},
{
"version": "1.49.1",
"date": "2026-10-07",
"body": "### archied\n\n- When the dashboard session expires, the page returns to the sign-in screen instead of showing errors on every page until a manual refresh."
},
{
"version": "1.49.0",
"date": "2026-10-07",
"body": "### archied\n\n- Extensions that declare no egress hosts (vault, email, gitea) reach the network again; only packages that declare their hosts are confined to them. The Accept dialog shows \"network: unrestricted\" for the rest.\n- The update check finds releases itself, and reuses a found release for ten minutes so repeated checks stay under GitHub's rate limit."
},
{
"version": "1.48.1",
"date": "2026-10-07",
"body": "### archied\n\n- **Settings → Status → Updates** now chooses only the release channel (stable, prereleases, or a pinned version). Update checks and installs always run the `archie-update-*` scripts installed beside the binaries; the dashboard can no longer set the command that runs."
},
{
"version": "1.48.0",
"date": "2026-10-07",
"body": "### archied\n\n- **Settings → Status → Updates** edits the update check and install commands. They default to the `archie-update-*` scripts installed beside the binaries, so a zip install shows **Install** on the Versions card with nothing configured. The `update_check_command` and `update_install_command` keys under `[chat.telegram]` are no longer read; remove them from `config.toml`.\n- `archie-update-install` run from a shell now reads `~/.config/archie/env`, so the pre-update backup of the bundled database no longer fails with `no password supplied`. Run it with `--auto` to install the newest release."
},
{
"version": "1.47.3",
"date": "2026-10-07",
"body": "### archied\n\n- **Settings → Extensions**: **Accept** in the authority review now accepts the extension, and **Remove** removes it. Both previously closed the dialog without doing anything."
},
{
"version": "1.47.2",
"date": "2026-10-06",
+394
View File
@@ -0,0 +1,394 @@
# External agent harnesses
**Status:** Approved
**Scope:** the harness model and, in "Session contract", the duplex session
that the setup terminal and the run-scoped live view both use. Code cites
this file for both (`ui/src/harness/SetupTerminal.vue`,
`internal/infrastructure/postgres/queries/harness_secrets.sql`).
## Decision
A workflow stage can run on an external coding-agent CLI instead of the
built-in agent loop: Claude Code, Codex, Pi, OMP, GitHub Copilot CLI, or any
other CLI packaged as a Docker Sandbox Kit v3. Each authenticates with an API
key or the operator's subscription login, and can carry the operator's own
skills, plugins, subagents and instructions.
Two boundaries hold regardless of what the harness does:
- **The real credential never enters the container.** The container sees
sentinel values; archie's egress proxy substitutes the real credential on
outbound requests and performs OAuth refresh on the host.
- **Archie's workflow guarantees are enforced outside the harness.** Git,
protected paths, read-only stages, gates and budgets are checked by archie
around the process and against the worktree afterwards.
## Kits
A harness is packaged as a Kit v3 image: an OCI image whose
`vnd.docker.sandbox.kit.descriptor` manifest annotation declares what the
harness needs. Archie reads descriptors with the published
`github.com/docker/sandbox-kit-spec/v3/spec` package and does not maintain
its own descriptor format. Published Kits run unmodified.
Archie is a partial Kit runtime. It implements these capability types:
| Type | Archie behaviour |
| ------------------- | ----------------------------------------------------------------------------------------- |
| `credential@1` | Proxy-managed only, described below. |
| `network-policy@1` | Enforced by the egress proxy. |
| `volume@1` | A Docker volume keyed by WorkflowExecution, deleted with it. |
| `lifecycle@1` | Install and startup hooks run on container create. |
| `agent-sessions@1` | The headless `prompt` and `resume` invocations for the stage. |
| `agent-skills@1` | The operator's skills store (`skills_dir`), mounted read-only whatever the Kit asks. |
| `agent-context@1` | The workload's profile file beside the workspace, indexing each Kit's guidance. |
| `resources@1` | Container CPU and memory limits. |
Every other type fails closed: a required entry refuses the launch, an
optional one is skipped and recorded on the step. `credential@1` with
`passthrough: true` and any credential for a forge service are refused.
The operator's own setup (plugins, subagents, instructions, settings) is a
mixin Kit, published with the harness Kit as a Kit set: the set merges their
layers into one image. It is versioned image content, rewritten on every
create, never state a run can modify.
## Contract
A harness run is a second implementation of `agentexec.Runner`. It takes the
same `Request` and returns the same `Result`, so workflows, the daemon and
the State Store do not know which runner ran a stage.
`agent-sessions@1` supplies the invocation. Archie supplies, per CLI, an
output adapter that translates the CLI's JSON output stream into
`ToolCallReporter` calls, `Usage` and a session ID. A CLI with no adapter can
run stages with no capture tools, reporting no usage; its token budget falls
back to wall clock.
Workflow definition validation rejects, at load time, a stage whose needs the
selected harness cannot meet: capture tools without an adapter, or a gate
retry without a `resume` verb.
## Selection
A harness is an agent profile. `docs/prds/event-automation.md` defines a
profile as the execution environment a workflow selects; a harness profile
names one Kit pinned by digest, a workload Kit or a published Kit set, in
place of a container image. Archie runs that Kit's image and composes no Kits
itself; a bare mixin is refused. Everything else about profiles
holds: a workflow names its profile, a stage may name a different one, and a
profile holds no secrets.
`archie-agent` builds the stage's runner from the stage's profile: the
ai-sdk agent loop for an image profile, the harness runner for a Kit
profile. Image profiles, and every workflow already written against them,
are unchanged. A Kit profile named by no workflow is a validation error.
A stage naming a different profile than its workflow's is a separate run,
not a mid-container runner switch: a Kit task runs in one container
provisioned for one profile (Container layout), so a stage that needs a
different one is its own callee, started the way `workflow.call` already
starts one (`docs/prds/event-automation.md` "Workflows call workflows"),
with its own agent profile resolved off its own task row. This needs no
mechanism beyond `workflow.call` plus per-task profile resolution
(`daemon.pinTaskProfile`), both of which already treat every task's
profile independently of any caller's -- see
`TestWorkflowCallerConformance` (staterpc) for the callee's independent
`Workflow` field and `TestWorkflowCallGivesACalleeItsOwnProfile` (daemon)
for the Kit-vs-image runner selection that independence produces.
Profiles are org resources in the State Store, so a Kit profile belongs to
one org, is granted to its workspaces like any profile, and applies without
a restart.
## Organisations
`docs/prds/orgs-and-access.md` is the boundary; a harness adds nothing that
crosses it.
- A Kit profile, its credential bindings and its OAuth tokens belong to one
org. No record is shared between orgs.
- A harness run is a run like any other: it acts as the workflow's identity,
under the run credential issued at dispatch.
- The egress proxy injects a credential only when the run credential carries
the secret the binding names. A Kit that declares a service the identity is
not granted gets no credential: a required one refuses the launch, an
optional one is skipped.
- A Kit's network policy is what the Kit asks for; the identity's network
grants are what the org allows. The proxy enforces the intersection.
- Concurrency slots are counted per binding within its org.
## Credentials
A credential binding maps a Kit's `credential@1` `service` name to an org
secret, which the org grants to identities like any secret. The secret holds
one of:
- **API key:** a value in the org's secret store.
- **OAuth:** tokens captured by the setup terminal, stored in the State Store
under the at-rest envelope of `binding-secret-encryption.md` with its own
domain separator.
The egress proxy resolves the binding for each request against the run
credential. The container gets the Kit's sentinel in the named variable or
credential file. OAuth refresh happens at the proxy, which intercepts the
Kit's `tokenEndpoint`, so a refreshed token is written to the State Store and
never returned to the container.
Runs sharing an OAuth binding share its subscription's rate limit.
`max_concurrent` on the binding (default 1) caps its simultaneous runs. A
stage waiting for a slot is `pending`, not `running`.
## Session contract
A **session** is a duplex PTY into a Kit container the daemon owns. One
concept covers both targets; only the address, the lifetime and the
established egress phase differ.
| | Setup session | Run session (deferred) |
| --- | --- | --- |
| Address | `{org, profile}` | `{org, task, attempt}` |
| Container | ephemeral, from the profile's Kit | the task's running container |
| Purpose | run the CLI's own login so the proxy captures the tokens | watch and optionally take over |
| Lifetime | bounded by the session TTL; the container is removed on close | bounded by the task |
- **Transport.** The daemon serves `harness.v1.HarnessService`, a bidi gRPC
stream, and the dashboard implements its `HarnessTerminal` interface
(`internal/webui`) over that client; the browser's WebSocket route is
unchanged. The contract lives on the daemon because the daemon owns
containers: the dashboard process must not link `internal/container`,
`internal/app/archied` or `internal/agentexec`.
- **Authorisation.** Opening a session is the `update` action on the secret,
decided by the dashboard's authorizer before the route runs
(`internal/webui/authorize.go`). The dashboard then dials the daemon with
the harness service token. The daemon does not re-authorise a human; it
trusts the asserted `{org, profile}` exactly as the task-action RPC trusts
an asserted actor. Because that authority is a shell in a container holding
an org's credential, the harness listener requires `HARNESS_TOKEN` even on
loopback, unlike the State Store's loopback convention.
- **Egress phase.** Token interception is filtered by phase
(`oauth.go`'s `rule.runtime == atRun`), and `EnterRuntime` is one-way. A
setup session therefore follows a run's lifecycle exactly: install hooks at
the install phase, `EnterRuntime` before the shell. A credential whose
token endpoint must be captured declares `phase: runtime`; a credential
capturing at install phase is captured by the install hooks that run before
the switch.
- **Credentials.** Capture is gated: the proxy intercepts a token endpoint
only for a service the session is granted (`Proxy.granted`). A setup session
has no task, so the daemon registers a session grant: its proxy token
resolves to `{org, the Kit's declared credential services the org binds}`.
That is the intersection the `update`-on-secret authorisation covers. The
real token never enters the container: the response is sentinelized and
stored as the org's harness secret.
- **Container discipline.** The setup container runs a keepalive entrypoint
with an empty `Config.Env`; the PTY is an exec as the Kit's harness user
with the harness env and nothing inherited from any worker. No worktree is
mounted. It is acquired through the container pool, so it carries the
`archie-daemon` label, the concurrency cap and the max-uptime reaper; the
session TTL is the smaller bound. Closing revokes the egress session,
removes the sandbox network and the Kit volumes, then releases the
container.
- **Kit unavailable.** The daemon serves the contract only when it has a
container pool and a Kit launcher; otherwise there is no listener, and a
client's refused connection is gRPC's `Unavailable`. The dashboard's
existing failure surface is the honest degrade; `/api/harness`'s `terminal`
field means "a client is configured", not "the daemon can run a Kit".
- **Run target.** `RunTarget{task_id, attempt}` opens a read-only view of a
running task's sandbox; it ships as the peek. The daemon resolves the pair
itself, so the dashboard never learns a container name: the task store says
whether the task is `running` and which attempt is current, and the Kit
launcher names the container `kitrun.TaskContainer(task_id, attempt)` (one
container per attempt, removed when the run ends; volumes remain task-scoped
for retries). A finished or unknown task, a task that is not running, an
attempt other than the running one, or a container that is gone is
`NotFound` with the reason in the message; an unreachable task store or
container engine is `Unavailable`. The view follows the container's stdout
and stderr, including the harness output forwarded by the worker, carries
no stdin and no exec, and
ends when the task's container ends or the client leaves, so it confers no
authority over the container. A read-only tail of the existing per-attempt
logs was rejected: those refresh only on task state changes and are the
daemon's account, not the sandbox's.
Mid-stream failures are shown as a reason in the terminal before it closes.
- **Authorisation.** The dashboard route `GET /api/tasks/{id}/sandbox` is
decided as `read-logs` on the task before the handler runs, so a caller
sees a sandbox exactly when they may read that task's log. The daemon
checks that the task belongs to the asserted org before following output.
- **Grants.** The view takes no grant of its own. The run's egress session
(its proxy token, install then `EnterRuntime`) already belongs to the run
and is revoked by `Launcher.Release` when the task ends; the view reads
after that phase switch and dies with the container.
- **Credentials.** The sandbox holds sentinels, never real credentials:
API keys are `egress.Sentinel` (`credentials.go`), and an OAuth token set
is sentinelised in the response and swapped back by the proxy only on a
granted service's resource host in the matching phase (`oauth.go`,
`injectOAuth`). The proxy runs in the daemon outside the container, so a
shell in the container sees only sentinels.
- **Takeover (interactive PTY into a running task) is out of scope for this
bead.** It needs the dangerous-command authority (`archie-core-sod1`) to
gate it, and that authority is not wired for it: the harness listener has
no approval step; `ToolSetOptions.Approval` in `internal/agentexec/modelloop`
is never set by a caller, so task runs never build approval-required tools;
approvals reach a human only on the chat paths (dashboard ask, Telegram,
Gateway). Takeover would also need an audit record naming the actor, task,
attempt and open/close times, and revocation tied to task end (the exec is
closed with the container). Those follow the setup session's shape
(`harnesssession.Session`, `container.ExecPTY`) once the authority exists.
- **Audit.** Opening and closing a session and a successful capture are
recorded with the asserted org and profile. A run view's open and close are
logged with the task and attempt; it reads output only, so it has no
takeover record.
Each session is the only interactive path into a harness container.
## Egress
Only Kit profiles run on the egress path. An image profile keeps the network
access its identity is granted, because workflows such as a network
investigator need raw DNS, ICMP and arbitrary TCP that no HTTP proxy
carries.
- **Sandbox network.** Each Kit run gets its own internal Docker bridge on
which the host holds no address. Host services, including those listening
on every interface, are unreachable from it by any address.
- **Relay.** One relay container, `archie-agent relay`, sits on the sandbox
networks under a fixed alias and on the host-reachable network. It
forwards three fixed ports, to the proxy, NATS and the State Store, and can
reach nothing else. It holds no secrets.
- **Proxy.** The proxy runs inside `archied`, listening on the host gateway
of the host-reachable network. It terminates TLS with a per-daemon CA
installed into the container at create, identifies the container by a
per-run proxy token, and applies that run's network policy and credential
inject rules. Upstream dials refuse loopback, and refuse internal addresses
unless the policy names the exact IP, so an allowed name cannot resolve
back into the host.
## Container layout
A Kit task runs in one container, from the Kit composition's image, on the
task's sandbox network. The worker and every harness invocation share it;
they are separated by user, not by container, so a run's cost scales with
the run and not with how many agents it fans out to.
- **Worker.** Archie mounts its own `archie-agent` binary read-only and runs
it as the container's entrypoint, as root. It holds the task's NATS and
State Store credentials in its own environment and in root-only files,
runs the workflow, and serves `archie-agent mcp`. The Kit image's
entrypoint is not PID 1.
- **Harness.** Every invocation runs as the Kit's user (the image's `User`,
default uid 1000) with an explicit environment: the Kit's variables, the
proxy environment and the credential sentinels, and nothing inherited from
the worker. That user cannot read the worker's environment or files, or
trace it. Sibling invocations in one run share the user and the worktree.
- **Hooks.** `lifecycle@1` install and startup hooks run as the user each
declares, before the worker accepts work.
- **Mounts.** The task worktree at the Kit's workspace directory, owned by the
Kit's user, and the read-only skills store. Nothing else from the host.
- **Network.** The relay forwards three fixed ports on the sandbox network:
the egress proxy, NATS and the State Store. The worker reaches NATS and the
State Store with its credentials; a harness can open those ports but holds
no credential either accepts.
- **Volumes.** `volume@1` paths persist for the life of the
WorkflowExecution, which covers session resume across gate iterations and
retries. They are not shared between executions.
## Archie tools inside the harness
Capture tools, which are how a stage returns structured results, are served
by `archie-agent mcp`, a stdio MCP server registered through the Kit's MCP
configuration or the adapter's flags. It serves the stage's `CaptureTools`
and the central tools the agent profile's `AllowTools` permits, never the
`workspace` toolset. The runner's `ToolLimits` bound its results; the
harness's own built-in tools are outside archie's limits.
## Enforcement around the harness
| Property | Enforcement |
| -------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| Model never runs git | No forge credential can be bound. Archie records `HEAD` and every ref before the run; any change fails the step (`failed`, policy class). The workflow commits the stage's changes. |
| Protected paths | A diff touching a `Protection` path fails the step. |
| Read-only stage | Any worktree diff fails the step. |
| Gate | Archie runs the stage's gate after the harness exits. On failure it resumes the session with the gate output, up to the stage's iteration budget. |
| Budget | Archie kills the process on wall-clock timeout, or when streamed usage passes the token budget. |
| Cancellation | Cancelling the step kills the process group, then the container. |
| Egress | Only the proxy is reachable; the Kit's network policy decides the rest. |
Harness settings that deny git or protected paths are added where the CLI
supports them, to save wasted turns. They are never the enforcement.
## Execution tree
A harness run is one `agent` StepExecution. Subagents the harness spawns
internally are not recorded as children. Their usage is included in the
step's `tokens_used` when the adapter reports it.
## Agents never hold a harness
An Agent reaches a harness only by invoking a Workflow whose stage names it.
No Agent tool, chat tool or capability exposes a harness, its setup terminal
or its credential. The stage boundary is what carries the container, the git
and path checks, the gate and the budget; a harness outside a stage has none
of them.
## Out of scope
- Chat through a harness. The gateway chat loop stays on the built-in
runtime.
- `network-policy@2` HTTP method and path rules; a Kit requiring it is
refused.
- Recording the harness's internal subagents as StepExecutions.
- Cost in currency.
- Choosing a harness automatically per task.
## Plan
1. Descriptor resolution with the Kit spec package; capability support
table and fail-closed refusal.
2. Egress proxy: internal network, per-daemon CA, per-container token,
`network-policy@1`.
3. `credential@1`: API-key bindings, sentinel rendering, inject rules.
4. Harness `Runner`: container create with worktree, volumes, lifecycle
hooks and the mounted `archie-agent`; `agent-sessions@1` invocation;
worktree checks for git, protected paths and read-only; budget and
cancellation.
5. Claude Code output adapter; `archie-agent mcp`; gate-and-resume loop.
6. Kit profiles, runner selection from the stage's profile, and definition
validation.
7. OAuth bindings as org secrets: token-endpoint interception, encrypted
storage, per-binding concurrency; dashboard setup terminal.
8. Output adapters for Codex, Pi, OMP and GitHub Copilot CLI.
## Verification
- The same workflow definition completes on `builtin` and on the Claude Code
Kit with the same `Result` shape and captures.
- After `/login` in the setup terminal, the container's credential file holds
only sentinels, the State Store holds the encrypted token, and a task run
authenticates.
- A real API key or OAuth token never appears in the container environment,
filesystem, worktree, task log or event stream.
- A request to a host outside the Kit's runtime allow list fails, including a
direct connection that bypasses the proxy settings.
- A harness instructed to `git commit` fails its step, and the task branch is
unchanged.
- A harness edit to a protected test file fails the step in a TDD fix stage.
- A read-only feasibility stage that writes a file fails.
- A file a run writes into a Kit-managed path such as the CLI's settings file
is gone on the next execution.
- A Kit requiring an unsupported capability is refused at launch with the
capability named.
- A stage with a failing gate resumes the same session with the gate output,
then succeeds or exhausts its iterations.
- Two tasks sharing an OAuth binding with `max_concurrent = 1` run one at a
time; the second waits as `pending`.
- Stopping a run kills the harness process and removes the container.
- A harness invocation cannot read the worker's NATS or State Store
credential from its environment, its files or `/proc`.
- A run fanning out to many agent invocations uses one container.
- A Kit profile in org A cannot be selected, and its binding cannot be
resolved, by a workflow in org B.
- A harness run whose identity is not granted a Kit's required credential is
refused at launch; the proxy never injects a secret outside the run
credential.
- The firewall investigation example runs unchanged on an image profile with
direct network access.
+239
View File
@@ -0,0 +1,239 @@
# User authorization boundaries
Archie separates authentication (which person or service is calling) from
permission to act on a resource. This inventory follows the [OWASP
Authentication Cheat Sheet](https://cheatsheetseries.owasp.org/cheatsheets/Authentication_Cheat_Sheet.html)
and [Authorization Cheat Sheet](https://cheatsheetseries.owasp.org/cheatsheets/Authorization_Cheat_Sheet.html),
including its guidance on attributes and relationships. It describes concrete
enforcement points and their regression tests; it is not a compliance certification.
## Transport and authority
The browser uses HTTP/JSON and SSE to the UI service. The UI service calls
State Store, Gateway, Messaging and Harness through their gRPC contracts.
Only the State Store opens the task/control-plane/account database; Gateway
owns conversation storage. Browser HTTP handlers never open either database.
Instance service credentials are administrative authority. They are not user
sessions and must never reach a browser, model, or agent container. A service
forwarding a principal preserves attribution and tenant scope; forwarding alone
does not mean every internal RPC evaluates Cedar. Most internal RPCs trust the
authenticated service and enforce their own resource scope. The public HTTP
boundary evaluates Cedar before invoking them. Task credentials have a separate,
explicit RPC allowlist and cannot call account or identity administration.
Local service listeners are trusted process boundaries. Loopback is not a
substitute for authentication on a remotely reachable listener. OIDC credentials,
local browser sessions, personal API tokens, service credentials, and task grants
must remain distinct.
## Policy and relationship model
`internal/domain/access` defines actions, resources, principals and memberships.
`internal/infrastructure/access` assembles Cedar entities and evaluates the
instance → organization → workspace → object chain. A forbid wins; each applicable
level must permit the operation. Missing or invalid policy chains fail closed.
The default policies grant actions through scoped roles. Policies can also
restrict a named identity, workspace environment, owner, state, source address,
signature, run or step. HTTP task decisions now include the stored initiating identity and task state; binding decisions include the stored binding state. The HTTP source address is the actual connection peer, not an unverified forwarded header. Membership and resource attributes must come from stored
records or authenticated context, never a request's assertion about itself.
Cedar permission is not ownership of another person's private data. Conversation
history, passkeys, recovery codes and personal tokens are bound to the person.
An organization or instance admin role does not override those relationships.
Account creation uses `manage_members` policy rather than a hard-coded admin
role. Authenticating through OIDC does not automatically grant membership, link
accounts by unverified email, or create an administrator.
## Enforcement inventory
| Location | Enforcement | Regression evidence |
| --- | --- | --- |
| Browser credential gate: `webui/auth.go`, `accounts.go` | Account session or verified provider identity; shared dashboard token is not a person when accounts are enabled. No API data before authentication. | `TestAccountBoundary`, `TestAccountSessionFailureIsClosed` |
| Browser policy gate: `webui/authorize.go` | Resolve principal from stored identity; evaluate action against stored org/workspace and resource kind before handler. Missing chain refuses. | `TestHTTPPolicyCannotBeBypassed`, `TestAuthorizeRefusesWithoutAChain`, `TestAuthorizeRecord`, `TestTaskAuthorizationUsesStoredAttributes`, `TestAuthorizeCreateIn` |
| Instance-only HTTP surfaces: `webui/instance_only.go`, server route registration | Hide instance-owned configuration and diagnostics from another org; policy still applies first. | `TestInstanceOnlyHidesTheInstanceFromOtherOrgs`, `TestInstanceAdminOrgs` |
| First setup: `webui/accounts.go`, `postgres/accounts.go` | Initial organization, workspace, instance URL, admin identity and password commit together. Transaction lock closes bootstrap after first account; later requests require `manage_members`. | Disposable PostgreSQL smoke verification for bootstrap race and rollback; account policy tests below |
| Account creation: `webui/accounts.go`, `staterpc/accounts.go`, `postgres/accounts.go` | Cedar `manage_members` at HTTP and RPC; target org is authenticated context. New local accounts cannot request owner role. | `TestAccountManagementUsesPolicy`, `TestAccountBoundary` |
| Local passwords and sessions: `identity/password.go`, `postgres/accounts.go` | Argon2id, independent salts, length/common-password checks, generic login refusal, account throttling, session hashes and expiry, revocation on logout/password change/suspension. | `TestPasswordVerification`, `TestAccountBoundary`; disposable PostgreSQL session verification |
| Transport and CSRF: `webui/accounts.go`, `api_tasks.go` | Password/account mutations require HTTPS or localhost development, JSON, custom CSRF header and same-origin requests. Forwarded headers require explicit trusted-proxy configuration. | `TestAuthenticationTransport`, existing mutation security tests |
| Personal account routes: `webui/accounts.go` | Active signed-in person; principal supplied by server. Password change verifies current password. Sensitive authenticator changes require recent sign-in. No caller-supplied target identity. | `TestAccountBoundary` |
| Recovery: `postgres/account_security.go`, `webui/accounts.go` | Fresh authenticated owner generates a hashed, one-use recovery secret. Redemption replaces password and revokes all sessions atomically. No email-existence disclosure. | Disposable PostgreSQL recovery verification |
| Passkeys: `postgres/passkeys.go`, `webui/passkeys.go` | Configured RP origin, browser-bound one-use expiring challenge, verified credential signature and user verification. Enrollment/management belong to caller. Credential counters persist under a row lock. | Disposable browser virtual-authenticator verification; task-grant denial tests |
| OIDC: `infrastructure/oidc`, `archieui/account_oidc.go`, `webui/auth.go` | Discovery and signed ID token with issuer/client audience/expiry, PKCE, state and nonce. Verified subject resolves to active identity. Browser-local session cannot outlive provider token, and provider credential remains checked. | `TestBrowserFlowVerifiesIDToken`, `TestProviderOutageIsUnavailableThenRecovers`, subject-claim tests; no external provider acceptance implied |
| OIDC settings: `webui/accounts.go`, `postgres/authentication.go` | Instance ownership plus policy; recent sign-in for changes. Public URL cannot change while enrolled passkeys would be orphaned. | `TestAccountBoundary`, `TestAuthenticationSettingsRejectUntrustedOrigins` |
| Org membership/workspaces/agent assignment: `webui/api_orgs.go`, `staterpc/orgs.go`, `postgres/orgs.go` | Own org or instance-owner relationship, allowed role vocabulary, membership escalation/last-owner guards, tenant-scoped SQL. | `TestOrgRPCsActInTheCallersOrg`, `TestSetMembershipRefusesUnknownRole`, org HTTP tests |
| Personal API tokens: `webui/api_personal_tokens.go`, `staterpc/personal_tokens.go`, `postgres/personal_tokens.go` | Token owner comes from principal; store only token hash; revoke only caller's binding. | `TestPersonalTokensWithoutPerson`, `TestPersonalTokenOwner` |
| gRPC caller attribution: `rpcidentity/identity.go` | Authenticated service may forward a principal; State Store resolves memberships/org itself. Task grant metadata is stripped. | `TestCallerAttribution` |
| Agent/task gRPC: `staterpc/task_grants.go` | Durable task-scoped grant, explicit allowed RPC and target task, no definition-pin replacement, no account RPCs. | `TestTaskCredentialsCannotManageAccounts`, `TestTaskCredentialCannotReplaceItsPin`, `TestRecordUsageAuthority`, `TestWorkerPresenceAuthority` |
| Chat HTTP: `webui/api_chat.go` | Server stamps sender and person thread; lists/read/turns/cancel/persona verify session ownership. Browser cannot submit another sender identity. | `TestWebChatOwnership`, `TestWebChatSenderCannotBeSuppliedByBrowser`; HTTP deny matrix |
| Chat history, slash commands and model session tools: `gateway/session_store_postgres.go` | Store filters `Get`, `List`, `GetByChannel`; shared read/write guards check person ownership as well as org. A saved session cannot change its conversation source to claim another person's history. Service curation runs without a person principal. | `TestSessionVisibility`; PostgreSQL smoke for history isolation |
| Chat questions/approvals: `webui/api_chat_ask.go` | Waiting question is bound to the initiating person, in addition to its unguessable ID. | `TestChatQuestionBelongsToInitiatingPerson` |
| Chat downloads: `webui/api_chat_files.go` | Offered file token expires and is bound to the initiating person; requests cannot supply a filesystem path. | `TestChatFileDownload` |
| Per-person memory: `archiegateway/chat_identity.go`, gateway memory tools, `sessioncurator` | Verified web sender identity scopes reads/writes. Curator refuses unidentified or mixed participants instead of guessing a person. | Gateway memory and session-curator tests |
| Workflow/task actions: `webui/api_tasks.go`, `domain/taskactions`, `staterpc` | Policy and stored workspace before action; domain transition, authority and durable dispatch guards remain in force. | HTTP deny matrix, task action/domain tests, task grant tests |
| Incoming webhooks: `forge/webhook`, `captureintake`, access delivery engine | Verify HMAC before parsing; derive source/org from stored source; evaluate source delivery policies and egress rules. No browser-user credential is inferred. | Webhook/capture tests, `TestAuthorizeDelivery` |
| Control-plane configuration, secrets, harness and extensions | Browser policy and instance-only restrictions where registered; service contracts carry principal/org; task grants cannot invoke administrative RPCs. | HTTP deny matrix, `TestAdministrationRoutes`, `TestChannelReloadRoute`, `TestConfineHarnessOrg` |
| Live HTTP/SSE and diagnostics | Authenticated policy gate applies before stream creation; org-scoped event reads; instance logs require administrative visibility. | HTTP deny matrix, instance-only tests, `TestServiceLogsRequireInstanceCredential` |
| Background schedules, dispatch, curation | Service/agent authority, not an implicit browser user. Task-scoped grants and org scopes remain distinct from initiating human attribution. | Existing dispatch, grant, memory and org-scope tests |
The policy HTTP matrix enumerates each registered protected route. It checks
both anonymous refusal and refusal by an explicit Cedar forbid, using the real
handler stack. Separate tests exercise allowed requests and personal ownership;
a denied-request matrix alone is not proof that a workflow works end to end.
## Public and personal routes
Public protocol entry points are `GET /api/auth/status`, `POST /api/auth/setup`,
`POST /api/auth/login`, `POST /api/auth/logout`, `POST /api/auth/recover`,
`POST /api/auth/passkey/begin`, `POST /api/auth/passkey/finish`,
`GET /oauth2/login`, `GET /oauth2/callback`, health probes and signed webhook
intake. Each has its own protocol checks. Static embedded app assets are public;
they do not contain user data. `/health/detailed` remains protected.
Personal routes under `/api/account/` require an active signed-in person and
server-derived ownership: password change, recovery-secret generation, passkey
begin/finish/list/delete. They do not require an organization administrator role.
`POST /api/accounts` and `/api/authentication` are administrative policy routes.
## HTTP route inventory
The following routes are covered by `TestHTTPPolicyCannotBeBypassed` in
`internal/webui/authorization_surfaces_test.go`. `authorize.go` maps each to its
action/resource; the handler may impose a stricter record or instance boundary.
| Method | Route example |
| --- | --- |
| DELETE | `/api/access/policies` |
| DELETE | `/api/bindings/7` |
| DELETE | `/api/event-types/7` |
| DELETE | `/api/extensions/sample` |
| DELETE | `/api/mappings/7` |
| DELETE | `/api/orgs/7/members/sample` |
| DELETE | `/api/sources/sample` |
| DELETE | `/api/tokens/7` |
| GET | `/api/access/policies` |
| GET | `/api/alerts` |
| GET | `/api/authentication` |
| GET | `/api/bindings` |
| GET | `/api/bindings/7` |
| GET | `/api/bindings/7/dispatches` |
| GET | `/api/capabilities` |
| GET | `/api/captures` |
| GET | `/api/captures/7/dispatches` |
| GET | `/api/channels` |
| GET | `/api/chat/file/sample` |
| GET | `/api/chat/sessions` |
| GET | `/api/chat/sessions/7/messages` |
| GET | `/api/chat/sessions/7/turns` |
| GET | `/api/chat/update` |
| GET | `/api/config` |
| GET | `/api/control-plane/apply-status` |
| GET | `/api/control-plane/audit` |
| GET | `/api/control-plane/catalog` |
| GET | `/api/control-plane/resources/sample` |
| GET | `/api/control-plane/resources/sample/history` |
| GET | `/api/curators` |
| GET | `/api/event-types` |
| GET | `/api/extensions` |
| GET | `/api/extensions/catalogue` |
| GET | `/api/harness/bindings` |
| GET | `/api/harness/terminal` |
| GET | `/api/identities` |
| GET | `/api/logs` |
| GET | `/api/mappings` |
| GET | `/api/mappings/7` |
| GET | `/api/orgs` |
| GET | `/api/orgs/7` |
| GET | `/api/orgs/7/members` |
| GET | `/api/orgs/7/workspaces` |
| GET | `/api/secret-engines` |
| GET | `/api/services` |
| GET | `/api/setup` |
| GET | `/api/skills` |
| GET | `/api/sources` |
| GET | `/api/stream` |
| GET | `/api/summary` |
| GET | `/api/task-meta` |
| GET | `/api/tasks` |
| GET | `/api/tasks/7` |
| GET | `/api/tasks/7/attempts` |
| GET | `/api/tasks/7/changes` |
| GET | `/api/tasks/7/debug` |
| GET | `/api/tasks/7/dispatches` |
| GET | `/api/tasks/7/logs` |
| GET | `/api/tasks/7/logs/download` |
| GET | `/api/tasks/7/sandbox` |
| GET | `/api/tokens` |
| GET | `/api/version` |
| GET | `/api/workflows` |
| PATCH | `/api/bindings/7` |
| PATCH | `/api/mappings/7` |
| POST | `/api/accounts` |
| POST | `/api/alerts/7/dismiss` |
| POST | `/api/bindings` |
| POST | `/api/bindings/7/approve` |
| POST | `/api/bindings/7/pause` |
| POST | `/api/bindings/7/resume` |
| POST | `/api/channels/7/reload` |
| POST | `/api/chat/answer` |
| POST | `/api/chat/cancel` |
| POST | `/api/chat/message` |
| POST | `/api/chat/persona` |
| POST | `/api/chat/stream` |
| POST | `/api/chat/update/defer` |
| POST | `/api/chat/update/install` |
| POST | `/api/control-plane/resources/sample/commands/sample` |
| POST | `/api/event-types` |
| POST | `/api/extensions` |
| POST | `/api/extensions/catalogue/sample/install` |
| POST | `/api/extensions/sample/accept` |
| POST | `/api/extensions/sample/approve-update` |
| POST | `/api/extensions/sample/rollback` |
| POST | `/api/extensions/sample/update` |
| POST | `/api/identities` |
| POST | `/api/identities/7/sample` |
| POST | `/api/mappings` |
| POST | `/api/mappings/preview` |
| POST | `/api/orgs` |
| POST | `/api/orgs/7/members` |
| POST | `/api/orgs/7/workspaces` |
| POST | `/api/secret-references/check` |
| POST | `/api/sources` |
| POST | `/api/sources/sample/approve-unsigned` |
| POST | `/api/sources/sample/secret` |
| POST | `/api/sources/sample/signing` |
| POST | `/api/tasks/7/action` |
| POST | `/api/tokens` |
| POST | `/api/work-requests` |
| PUT | `/api/access/policies` |
| PUT | `/api/authentication` |
| PUT | `/api/event-types/7` |
| PUT | `/api/extensions/sample/enabled` |
| PUT | `/api/extensions/sample/update-policy` |
| PUT | `/api/orgs/7/agents/sample` |
| PUT | `/api/orgs/7/members/sample` |
| PUT | `/api/sources/sample/delivery-header` |
| PUT | `/api/sources/sample/name` |
| PUT | `/api/workflows/7/enabled` |
## Verification limits
A synthetic browser authenticator verifies protocol behavior, not a physical
security key or every operating system. An OIDC unit test is not acceptance
against a deployed provider. Disposable PostgreSQL smoke checks verify real SQL
without putting database-dependent tests in the fast unit suite.
Existing anonymous conversations are retained but are not silently assigned to
a new person. Existing service credentials remain privileged; this inventory
does not claim that all internal gRPC methods have independent Cedar gates.
When changing a user-facing operation, update its inventory entry and its
allow/deny behavior cases. Include another user in the same organization, a
caller in another organization, a missing principal, unavailable policy storage,
and the applicable owner/workspace/instance restrictions.
Bootstrap is unauthenticated until the first administrator is created. Complete
setup before exposing a new installation to untrusted users. Last-active-owner
protection for identity suspension/retirement is tracked separately as
`archie-core-d7avc`; membership-policy guards are not that lifecycle guard.
+2
View File
@@ -0,0 +1,2 @@
node_modules/
out/
+61
View File
@@ -0,0 +1,61 @@
# e2e
Validation of the whole stack, run by hand or by an agent, never by
`task check` or CI. It needs PostgreSQL 18, five services, Vite and a browser,
which the gate must not. Docker is not needed: without it tasks queue and
park, which is useful data for the dashboard.
| File | What it does |
| --- | --- |
| `stack.sh` | Starts PostgreSQL 18 (port 55432) and the five services as local processes, then Vite on 5173 |
| `seed.sh` | Seeds the stack through the dashboard API; safe to rerun |
| `run.cjs` | Runs `checks/*.cjs` in headless Chromium; screenshots to `out/` |
| `checks/` | One module per behaviour |
| `results/` | One committed report per release |
## Run
```bash
task e2e:up # build, start everything, seed
task e2e:check # all checks; or: node e2e/run.cjs routes
task e2e:down # stop; e2e/stack.sh reset also deletes the data
```
Nothing reads the installed config: `stack.sh` points the XDG directories at
`${ARCHIE_DEV_DIR:-~/.local/state/archie-dev}` and writes its overlay from
`deployments/dev.toml`. Logs are in its `logs/`. `e2e/stack.sh restart archied`
restarts one service.
The first run needs `npm ci --prefix e2e` and `npx --prefix e2e playwright
install chromium`. In the cloud environment, Chromium and Playwright are
preinstalled; run with `NODE_PATH=$(npm root -g)`.
## Release evidence
```bash
node e2e/run.cjs --record v1.54.0
```
writes `results/v1.54.0.md`, the pass or fail of every check for that
release. Commit it with the release. Screenshots stay in `out/`.
## Writing a check
A module in `checks/` exports `async ({ page, go, shot, fail })`. `go(path)`
loads a dashboard route, `shot(label)` saves a screenshot, `fail(msg)` records
a problem without stopping. A throw, a `fail`, or any console or page error
fails the check. Wait for elements, never for `networkidle`: the event stream
never goes idle.
`routes` loads every static route and flags text with stray leading,
trailing or doubled spaces. `seeded` checks the seed is visible.
## Pitfalls
| Symptom | Cause | Fix |
| --- | --- | --- |
| State Store exits: server is older than PostgreSQL 18 | another server answered | `stack.sh` uses port 55432; set `ARCHIE_DEV_PGPORT` if that is taken |
| Work request: repository is not configured | archied reads repository policies only at boot | `seed.sh` restarts it; do the same after changing policies |
| Your own shell is killed (exit 144) | `pkill -f` matched its own command line | `stack.sh stop` or `restart`, which kill by PID |
| `npx shadcn-vue add` cannot fetch the registry | Node's fetch ignores the cloud agent proxy | `curl` the registry JSON and copy the files |
| Postgres aborts in the cloud | `/tmp/claude-0` is reset to mode 700 | `stack.sh` keeps the root cluster in `/var/lib/archie-dev` |
+36
View File
@@ -0,0 +1,36 @@
// Every static route loads without errors, renders, and shows no text with
// stray leading, trailing or doubled spaces.
const fs = require('fs');
const path = require('path');
function routes() {
const src = fs.readFileSync(path.join(__dirname, '../../ui/src/router/index.ts'), 'utf8');
const out = [];
for (const line of src.split('\n')) {
if (/redirect:/.test(line)) continue;
const m = line.match(/path: "([^"]*)"/);
if (!m || m[1].includes(':')) continue;
if (m[1].startsWith('/')) out.push(m[1]);
else if (m[1]) out.push('/org/' + m[1]);
}
return [...new Set(out)];
}
module.exports = async ({ page, go, shot, fail }) => {
const stray = new Map();
for (const r of routes()) {
await go(r);
await shot(r);
// Log lines are data, not copy.
if (r === '/logs') continue;
const bad = await page.evaluate(() => [...document.querySelectorAll('body *')]
.filter((el) => !el.children.length && !el.closest('pre,code,textarea,.xterm') && el.getClientRects().length && !getComputedStyle(el).whiteSpace.startsWith('pre'))
.map((el) => el.textContent)
.filter((t) => t.trim() && (t !== t.trim() || / /.test(t.trim()))));
for (const t of bad) {
const key = JSON.stringify(t.length > 60 ? t.slice(0, 60) + '…' : t);
stray.set(key, (stray.get(key) || new Set()).add(r));
}
}
for (const [text, where] of stray) fail(`stray whitespace ${text} on ${[...where].join(', ')}`);
};
+9
View File
@@ -0,0 +1,9 @@
// The seed (e2e/seed.sh) is visible where an operator would look for it.
module.exports = async ({ page, go, shot }) => {
await go('/tasks');
await page.getByText('Fix flaky login test').first().waitFor({ timeout: 10000 });
await shot('tasks');
await go('/events');
await page.getByText('e2e-hook').first().waitFor({ timeout: 10000 });
await shot('events');
};
+42
View File
@@ -0,0 +1,42 @@
{
"name": "archie-e2e",
"lockfileVersion": 3,
"requires": true,
"packages": {
"": {
"name": "archie-e2e",
"devDependencies": {
"playwright": "1.64.0"
}
},
"node_modules/playwright": {
"version": "1.64.0",
"resolved": "https://registry.npmjs.org/playwright/-/playwright-1.64.0.tgz",
"integrity": "sha512-kVzTnYFYEyQ9RGzBjNCd2+yEqWmVbxl3AHPynv3qQ6JTjv4a+kwRcnqQwIpuP3NnrnIrE/Q6wWAIAvMfIE2/Yg==",
"dev": true,
"license": "Apache-2.0",
"dependencies": {
"playwright-core": "1.64.0"
},
"bin": {
"playwright": "cli.js"
},
"engines": {
"node": ">=20"
}
},
"node_modules/playwright-core": {
"version": "1.64.0",
"resolved": "https://registry.npmjs.org/playwright-core/-/playwright-core-1.64.0.tgz",
"integrity": "sha512-T9r+MZkTECl2+oUcZ26YgZzTrYidTsIhYYkJix+6iDOymOb1FB2RLLSOxRGnZGv5DStt+aduIVbt+E8WygXYpg==",
"dev": true,
"license": "Apache-2.0",
"bin": {
"playwright-core": "cli.js"
},
"engines": {
"node": ">=20"
}
}
}
}
+11
View File
@@ -0,0 +1,11 @@
{
"name": "archie-e2e",
"private": true,
"description": "Validation of a running local stack; never part of the gate",
"scripts": {
"check": "node run.cjs"
},
"devDependencies": {
"playwright": "1.64.0"
}
}
+62
View File
@@ -0,0 +1,62 @@
// Drive the dashboard of a running local stack (e2e/stack.sh) in headless
// Chromium. Each module in checks/ exports async ({ page, go, shot, fail }).
// A check fails on a thrown error or any console or page error.
//
// node e2e/run.cjs [--record VERSION] [check ...]
//
// Screenshots and results.json go to e2e/out/<stamp>/. --record also writes
// e2e/results/VERSION.md, the committed evidence for a release.
const fs = require('fs');
const path = require('path');
const { chromium } = require('playwright');
const base = process.env.ARCHIE_E2E_URL || 'http://localhost:5173';
const args = process.argv.slice(2);
let record = '';
if (args[0] === '--record') { record = args[1]; args.splice(0, 2); }
const names = args.length ? args : fs.readdirSync(path.join(__dirname, 'checks')).filter((f) => f.endsWith('.cjs')).map((f) => f.slice(0, -4)).sort();
const stamp = new Date().toISOString().replace(/[:.]/g, '-');
const out = path.join(__dirname, 'out', stamp);
fs.mkdirSync(out, { recursive: true });
(async () => {
const browser = await chromium.launch();
const results = [];
for (const name of names) {
const check = require(path.join(__dirname, 'checks', name + '.cjs'));
// A fresh page per check: open event streams pile up across navigations.
const page = await browser.newPage({ viewport: { width: 1440, height: 900 } });
const errors = [];
const notes = [];
page.on('pageerror', (e) => errors.push(String(e)));
page.on('console', (m) => m.type() === 'error' && errors.push(m.text()));
// Wait for load, never networkidle: the event stream never goes idle.
const go = async (p) => { await page.goto(base + p, { waitUntil: 'load' }); await page.waitForTimeout(1200); };
const shot = (label) => page.screenshot({ path: path.join(out, `${name}-${label.replace(/[^a-z0-9-]+/gi, '_') || 'root'}.png`) });
const fail = (msg) => notes.push(msg);
const started = Date.now();
let error = '';
try { await check({ page, go, shot, fail }); } catch (e) { error = String(e && e.message || e); }
await page.close();
const problems = [error, ...notes, ...errors].filter(Boolean);
results.push({ check: name, ok: problems.length === 0, ms: Date.now() - started, problems });
console.log(`${problems.length ? 'FAIL' : 'ok '} ${name}${problems.length ? '\n ' + problems.join('\n ') : ''}`);
}
await browser.close();
fs.writeFileSync(path.join(out, 'results.json'), JSON.stringify(results, null, 2));
if (record) writeRecord(record, results);
console.log(`evidence: ${out}`);
process.exit(results.every((r) => r.ok) ? 0 : 1);
})().catch((e) => { console.error(e); process.exit(2); });
function writeRecord(version, results) {
const dir = path.join(__dirname, 'results');
fs.mkdirSync(dir, { recursive: true });
const rows = results.map((r) => `| ${r.check} | ${r.ok ? 'pass' : 'fail'} | ${r.problems.join('<br>').replace(/\|/g, '\\|') || ''} |`);
const passed = results.filter((r) => r.ok).length;
fs.writeFileSync(path.join(dir, `${version}.md`), [
`# ${version}`, '',
`Run ${new Date().toISOString()} against ${base}: ${passed} of ${results.length} checks passed.`, '',
'| Check | Result | Problems |', '| --- | --- | --- |', ...rows, '',
].join('\n'));
}
Executable
+50
View File
@@ -0,0 +1,50 @@
#!/usr/bin/env bash
# Seed a running local stack (e2e/stack.sh) through the dashboard's own HTTP
# API, so the seed exercises real validation. Safe to run again: each step
# skips what already exists.
set -euo pipefail
base=${ARCHIE_E2E_API:-http://localhost:8484}
here=$(cd "$(dirname "$0")" && pwd)
H=(-H 'Content-Type: application/json' -H 'X-Archie-CSRF: 1' -H "Origin: $base")
get() { curl -fsS "$base$1"; }
post() { curl -fsS -X POST "$base$1" "${H[@]}" -d "$2"; }
# jq-free JSON reads: py EXPR, with the parsed body as d.
py() { python3 -c "import sys, json; d = json.load(sys.stdin); print($1)"; }
# Repository policy. archied applies it live; the pause lets the watch publish it.
if ! get /api/control-plane/resources/repository-policies | grep -q '"widgets"'; then
version=$(get /api/control-plane/resources/repository-policies | py 'd["resource"]["version"]')
post /api/control-plane/resources/repository-policies/commands/replace \
"{\"value\":[{\"owner\":\"acme\",\"name\":\"widgets\",\"base\":\"main\"}],\"expected_version\":$version}" >/dev/null
sleep 5
fi
# Work requests. Without Docker they park within a minute, which fills
# Needs you, Quality gates and the task pages.
tasks=$(get /api/tasks)
for t in "Fix flaky login test" "Add retry to webhook sender" "Document the gate" "Speed up the CI cache"; do
grep -qF "\"$t\"" <<<"$tasks" && continue
post /api/work-requests "{\"identity\":\"archie-dev\",\"repository\":\"acme/widgets\",\"workflow\":\"triage\",\"title\":\"$t\",\"instructions\":\"Do it.\"}" >/dev/null
done
# An unsigned capture source with one delivery, an event type, a mapping and
# a binding left pending approval.
src=e2e-hook
if ! get /api/sources | grep -q "\"$src\""; then
post /api/sources "{\"path\":\"$src\"}" >/dev/null
post "/api/sources/$src/signing" '{"signed":false}' >/dev/null
post "/api/sources/$src/approve-unsigned" '{}' >/dev/null
bash "$here/../tools/webhook-send.sh" --url "$base" --source "$src" --kind github-issue >/dev/null
fi
et=$(get /api/event-types | py '"".join(t["id"] for t in (d.get("event_types", []) if isinstance(d, dict) else d) if t["name"] == "issue-opened")')
if [ -z "$et" ]; then
et=$(post /api/event-types "{\"source\":\"$src\",\"name\":\"issue-opened\",\"rule\":{\"headers\":[],\"payload\":[{\"path\":\"action\",\"op\":\"equals\",\"value\":\"opened\"}]},\"example\":{\"headers\":{},\"body\":\"{\\\"action\\\":\\\"opened\\\",\\\"issue\\\":{\\\"title\\\":\\\"x\\\"}}\"}}" | py 'd["id"]')
fi
if ! get /api/bindings | grep -q '"e2e-binding"'; then
mapping=$(post /api/mappings "{\"name\":\"e2e-issue\",\"source_hint\":\"$src\",\"event_type_id\":\"$et\",\"fields\":[{\"name\":\"title\",\"path\":\"issue.title\",\"type\":\"string\",\"required\":true}]}" | py 'd["id"]')
post /api/bindings "{\"name\":\"e2e-binding\",\"matcher\":{\"source\":\"$src\"},\"mapping_id\":\"$mapping\",\"workflow\":\"feasibility\",\"owner\":\"acme\",\"repo\":\"widgets\"}" >/dev/null
fi
echo "seeded: $(get /api/tasks | py 'len(d.get("tasks", d) if isinstance(d, dict) else d)') tasks"
Executable
+140
View File
@@ -0,0 +1,140 @@
#!/usr/bin/env bash
# Run the whole stack as local processes against a throwaway PostgreSQL 18,
# for manual and scripted checks. Never part of the gate. Nothing here reads
# the installed config: XDG paths point into the dev directory.
#
# e2e/stack.sh start | stop | restart <service> | status | reset
set -euo pipefail
repo=$(cd "$(dirname "$0")/.." && pwd)
dir=${ARCHIE_DEV_DIR:-${XDG_STATE_HOME:-$HOME/.local/state}/archie-dev}
pgport=${ARCHIE_DEV_PGPORT:-55432}
services=(archie-state-store archie-gateway archied archie-messaging archie-ui)
# initdb refuses root, and postgres cannot traverse root's home.
if [ "$(id -u)" -eq 0 ]; then
pgdata=/var/lib/archie-dev/pg
as_pg() { runuser -u postgres -- "$@"; }
else
pgdata=$dir/pg
as_pg() { "$@"; }
fi
export XDG_CONFIG_HOME=$dir/config XDG_DATA_HOME=$dir/data XDG_STATE_HOME=$dir/state
pgbin() {
local d
for d in /usr/lib/postgresql/18/bin "$(dirname "$(command -v postgres 2>/dev/null || echo /x/x)")"; do
if [ -x "$d/postgres" ] && "$d/postgres" --version | grep -q ' 18\.'; then
echo "$d"
return
fi
done
echo "PostgreSQL 18 server not found" >&2
exit 1
}
wait_port() {
for _ in $(seq 1 "${3:-60}"); do
(echo >"/dev/tcp/127.0.0.1/$1") 2>/dev/null && return 0
sleep 1
done
echo "$2 did not listen on $1; see $dir/logs/$2.log" >&2
return 1
}
start_pg() {
local bin
bin=$(pgbin)
if [ ! -f "$pgdata/PG_VERSION" ]; then
mkdir -p "$pgdata"
[ "$(id -u)" -ne 0 ] || chown postgres "$pgdata"
as_pg "$bin/initdb" -D "$pgdata" -U archie --auth=trust >/dev/null
fi
if ! as_pg "$bin/pg_ctl" -D "$pgdata" status >/dev/null 2>&1; then
as_pg "$bin/pg_ctl" -D "$pgdata" -w -l "$pgdata/server.log" \
-o "-p $pgport -k $pgdata -c listen_addresses=127.0.0.1" start >/dev/null
fi
"$bin/psql" -h 127.0.0.1 -p "$pgport" -U archie -d postgres -qtAc \
"SELECT 1 FROM pg_database WHERE datname = 'archie'" | grep -q 1 ||
"$bin/createdb" -h 127.0.0.1 -p "$pgport" -U archie archie
}
prepare() {
mkdir -p "$dir"/{bin,logs,pids} "$XDG_CONFIG_HOME/archie" "$XDG_DATA_HOME" "$XDG_STATE_HOME"
[ -f "$XDG_CONFIG_HOME/archie/config.toml" ] || cp "$repo/config.example.toml" "$XDG_CONFIG_HOME/archie/config.toml"
sed "s#^database_url = .*#database_url = \"postgres://archie@127.0.0.1:$pgport/archie?sslmode=disable\"#" \
"$repo/deployments/dev.toml" >"$dir/dev.toml"
local s
for s in "${services[@]}"; do
(cd "$repo" && go build -o "$dir/bin/$s" "./cmd/$s")
done
}
# spawn NAME DIR CMD...: a detached session leader, so stop takes its children.
spawn() {
local name=$1 cwd=$2
shift 2
setsid -f bash -c 'cd "$1" && echo $$ >"$2" && shift 2 && exec "$@"' _ "$cwd" "$dir/pids/$name" "$@" \
</dev/null >"$dir/logs/$name.log" 2>&1
}
start_service() {
spawn "$1" "$repo" "$dir/bin/$1" -config-overlay "$dir/dev.toml"
}
stop_pid() {
local f=$1 pid
[ -f "$f" ] || return 0
pid=$(cat "$f")
if kill -- "-$pid" 2>/dev/null; then
for _ in $(seq 1 20); do kill -0 "$pid" 2>/dev/null || break; sleep 0.5; done
kill -9 -- "-$pid" 2>/dev/null || true
fi
rm -f "$f"
}
start() {
prepare
start_pg
start_service archie-state-store
wait_port 9091 archie-state-store 90
local s
for s in archie-gateway archied archie-messaging archie-ui; do start_service "$s"; done
wait_port 8484 archie-ui
[ -d "$repo/ui/node_modules" ] || (cd "$repo/ui" && npm ci --no-audit --no-fund >/dev/null)
spawn vite "$repo/ui" npx vite --host 127.0.0.1 --port 5173 --strictPort
wait_port 5173 vite
echo "dashboard: http://localhost:5173 logs: $dir/logs"
}
stop() {
local f
for f in "$dir"/pids/*; do stop_pid "$f"; done
if [ -f "$pgdata/PG_VERSION" ]; then
as_pg "$(pgbin)/pg_ctl" -D "$pgdata" -m fast stop >/dev/null 2>&1 || true
fi
}
case ${1:-} in
start) start ;;
stop) stop ;;
restart)
stop_pid "$dir/pids/${2:?service}"
start_service "$2"
;;
status)
for f in "$dir"/pids/*; do
[ -f "$f" ] || continue
if kill -0 "$(cat "$f")" 2>/dev/null; then echo "up $(basename "$f")"; else echo "down $(basename "$f")"; fi
done
;;
reset)
stop
rm -rf "$dir" "$pgdata"
;;
*)
echo "usage: $0 start | stop | restart <service> | status | reset" >&2
exit 2
;;
esac
+46 -63
View File
@@ -1,130 +1,113 @@
module github.com/samcharles93/archie-core
go 1.27.0
go 1.27.2
require (
cel.dev/expr v0.25.3 // indirect
dario.cat/mergo v1.0.2 // indirect
github.com/Azure/go-ansiterm v0.0.0-20260917205352-e937bb47801a // indirect
github.com/Microsoft/go-winio v0.6.2 // indirect
github.com/ProtonMail/go-crypto v1.5.1 // indirect
github.com/Microsoft/go-winio v0.6.4 // indirect
github.com/ProtonMail/go-crypto v1.5.2 // indirect
github.com/antithesishq/antithesis-sdk-go v0.8.0 // indirect
github.com/antlr4-go/antlr/v4 v4.13.1 // indirect
github.com/cenkalti/backoff/v4 v4.3.0 // indirect
github.com/cespare/xxhash/v2 v2.3.0 // indirect
github.com/cloudflare/circl v1.6.5 // indirect
github.com/containerd/errdefs v1.0.0 // indirect
github.com/containerd/errdefs/pkg v0.3.0 // indirect
github.com/containerd/log v0.2.0 // indirect
github.com/containerd/platforms v1.0.0-rc.5 // indirect
github.com/cpuguy83/dockercfg v0.3.2 // indirect
github.com/distribution/reference v0.6.0 // indirect
github.com/docker/go-connections v0.8.1 // indirect
github.com/docker/go-connections v0.8.2 // indirect
github.com/docker/go-units v0.5.0 // indirect
github.com/dustin/go-humanize v1.1.0 // indirect
github.com/ebitengine/purego v0.11.1 // indirect
github.com/emirpasic/gods v1.18.1 // indirect
github.com/fatih/color v1.19.0 // indirect
github.com/felixge/httpsnoop v1.1.0 // indirect
github.com/fxamacker/cbor/v2 v2.9.4 // indirect
github.com/go-git/gcfg/v2 v2.0.2 // indirect
github.com/go-jose/go-jose/v4 v4.1.5 // indirect
github.com/go-json-experiment/json v0.0.0-20260820222146-c27c302e5fc3 // indirect
github.com/go-logr/logr v1.4.4 // indirect
github.com/go-logr/stdr v1.2.2 // indirect
github.com/go-ole/go-ole v1.3.0 // indirect
github.com/go-viper/mapstructure/v2 v2.5.0 // indirect
github.com/go-webauthn/x v0.3.1 // indirect
github.com/golang/protobuf v1.5.4 // indirect
github.com/google/go-tpm v0.9.8 // indirect
github.com/google/jsonschema-go v0.4.3 // indirect
github.com/hashicorp/yamux v0.1.2 // indirect
github.com/jackc/pgpassfile v1.0.0 // indirect
github.com/jackc/pgservicefile v0.0.0-20240606120523-5a60cdf6a761 // indirect
github.com/jackc/puddle/v2 v2.2.2 // indirect
github.com/jackc/puddle/v2 v2.2.3 // indirect
github.com/kevinburke/ssh_config v1.6.0 // indirect
github.com/klauspost/compress v1.20.0 // indirect
github.com/klauspost/cpuid/v2 v2.4.0 // indirect
github.com/lufia/plan9stats v0.0.0-20260802145828-341c2f0c90b5 // indirect
github.com/magiconair/properties v1.18.12 // indirect
github.com/mattn/go-colorable v0.1.14 // indirect
github.com/klauspost/compress v1.20.3 // indirect
github.com/mattn/go-colorable v0.1.16 // indirect
github.com/mattn/go-isatty v0.0.24 // indirect
github.com/mfridman/interpolate v0.0.2 // indirect
github.com/minio/highwayhash v1.0.4 // indirect
github.com/moby/docker-image-spec v1.3.1 // indirect
github.com/moby/go-archive v0.3.3 // indirect
github.com/moby/patternmatcher v0.6.1 // indirect
github.com/moby/sys/sequential v0.7.0 // indirect
github.com/moby/sys/user v0.4.1 // indirect
github.com/moby/sys/userns v0.2.1 // indirect
github.com/moby/term v0.5.2 // indirect
github.com/nats-io/jwt/v2 v2.8.2 // indirect
github.com/nats-io/nkeys v0.4.16 // indirect
github.com/nats-io/nuid v1.0.1 // indirect
github.com/oklog/run v1.1.0 // indirect
github.com/pjbgf/sha1cd v0.6.0 // indirect
github.com/oklog/run v1.2.0 // indirect
github.com/philhofer/fwd v1.2.0 // indirect
github.com/pjbgf/sha1cd v0.7.0 // indirect
github.com/pkg/browser v0.0.0-20240102092130-5ac0b6a4141c // indirect
github.com/power-devops/perfstat v0.0.0-20260916203055-22a1a467d9f0 // indirect
github.com/segmentio/asm v1.2.1 // indirect
github.com/segmentio/encoding v0.5.4 // indirect
github.com/sergi/go-diff v1.4.0 // indirect
github.com/sethvargo/go-retry v0.4.0 // indirect
github.com/shirou/gopsutil/v4 v4.26.8 // indirect
github.com/sethvargo/go-retry v0.5.0 // indirect
github.com/sirupsen/logrus v1.10.2 // indirect
github.com/stretchr/testify v1.12.1 // indirect
github.com/tklauser/go-sysconf v0.4.0 // indirect
github.com/tklauser/numcpus v0.12.0 // indirect
github.com/tinylib/msgp v1.6.4 // indirect
github.com/x448/float16 v0.8.4 // indirect
github.com/yosida95/uritemplate/v3 v3.0.2 // indirect
github.com/yusufpapurcu/wmi v1.2.4 // indirect
go.opentelemetry.io/auto/sdk v1.2.1 // indirect
go.opentelemetry.io/contrib/instrumentation/net/http/otelhttp v0.71.0 // indirect
go.opentelemetry.io/otel v1.46.0 // indirect
go.opentelemetry.io/otel/metric v1.46.0 // indirect
go.opentelemetry.io/otel/trace v1.46.0 // indirect
go.opentelemetry.io/contrib/instrumentation/net/http/otelhttp v0.72.0 // indirect
go.opentelemetry.io/otel v1.47.0 // indirect
go.opentelemetry.io/otel/log v1.47.0 // indirect
go.opentelemetry.io/otel/metric v1.47.0 // indirect
go.opentelemetry.io/otel/trace v1.47.0 // indirect
go.uber.org/multierr v1.11.0 // indirect
go.yaml.in/yaml/v3 v3.0.5 // indirect
golang.org/x/crypto v0.57.0 // indirect
golang.org/x/exp v0.0.0-20260908205506-85c1c2202aba // indirect
golang.org/x/sync v0.23.0 // indirect
golang.org/x/text v0.42.0 // indirect
golang.org/x/exp v0.0.0-20261013213113-2ab7f3cbc5fb // indirect
golang.org/x/sync v0.24.0 // indirect
golang.org/x/text v0.43.0 // indirect
golang.org/x/time v0.16.0 // indirect
google.golang.org/genproto/googleapis/api v0.0.0-20260921155816-b14227669459 // indirect
google.golang.org/genproto/googleapis/rpc v0.0.0-20260921155816-b14227669459 // indirect
modernc.org/libc v1.77.1 // indirect
modernc.org/sqlite v1.59.0 // indirect
google.golang.org/genproto/googleapis/api v0.0.0-20261012220534-4341edf3f941 // indirect
google.golang.org/genproto/googleapis/rpc v0.0.0-20261012220534-4341edf3f941 // indirect
)
require (
cel.dev/cel-go v0.32.0
github.com/BurntSushi/toml v1.6.0
github.com/cedar-policy/cedar-go v1.8.0
github.com/containerd/errdefs v1.0.0
github.com/coreos/go-oidc/v3 v3.21.0
github.com/docker/sandbox-kit-spec/v3 v3.0.0-m.6
github.com/go-git/go-billy/v6 v6.0.0-alpha.2
github.com/go-git/go-git/v6 v6.0.0-alpha.5
github.com/docker/sandbox-kit-spec/v3 v3.0.0-m.8
github.com/go-git/go-billy/v6 v6.0.0-beta.1
github.com/go-git/go-git/v6 v6.0.0-beta.1
github.com/go-jose/go-jose/v4 v4.1.5
github.com/go-telegram/bot v1.27.0
github.com/go-webauthn/webauthn v0.18.2
github.com/golang-jwt/jwt/v5 v5.3.1
github.com/google/jsonschema-go v0.4.3
github.com/google/uuid v1.6.0
github.com/hashicorp/go-hclog v1.6.3
github.com/hashicorp/go-plugin v1.8.0
github.com/jackc/pgx/v5 v5.11.0
github.com/moby/moby/api v1.56.0
github.com/moby/moby/client v0.6.0
github.com/moby/moby/api v1.56.1
github.com/moby/moby/client v0.6.2
github.com/modelcontextprotocol/go-sdk v1.8.0
github.com/nats-io/nats-server/v2 v2.15.0
github.com/nats-io/nats-server/v2 v2.15.1
github.com/nats-io/nats.go v1.54.0
github.com/opencontainers/go-digest v1.0.0
github.com/opencontainers/image-spec v1.1.1
github.com/pressly/goose/v3 v3.28.0
github.com/samcharles93/ai-sdk v0.1.42
github.com/testcontainers/testcontainers-go v0.44.0
github.com/testcontainers/testcontainers-go/modules/postgres v0.44.0
github.com/pressly/goose/v3 v3.29.0
github.com/samcharles93/ai-sdk v0.3.0
go.lsp.dev/jsonrpc2 v1.0.1
go.lsp.dev/protocol v1.0.1
go.lsp.dev/uri v1.0.1
golang.org/x/mod v0.41.0
golang.org/x/net v0.59.0
golang.org/x/crypto v0.58.0
golang.org/x/mod v0.42.0
golang.org/x/net v0.61.0
golang.org/x/oauth2 v0.37.0
golang.org/x/sys v0.48.0
golang.org/x/term v0.46.0
google.golang.org/grpc v1.85.0-dev.0.20260825072537-93e31b48545e
golang.org/x/sys v0.49.0
golang.org/x/term v0.47.0
google.golang.org/grpc v1.86.0-dev
google.golang.org/protobuf v1.36.12
gopkg.in/yaml.v3 v3.0.1
oras.land/oras-go/v2 v2.6.2
oras.land/oras-go/v2 v2.6.3
)
+98 -140
View File
@@ -2,18 +2,12 @@ cel.dev/cel-go v0.32.0 h1:irvpFKr5EuGPyxeME03ERh0rii1TX+BDAnB9eL3IvNk=
cel.dev/cel-go v0.32.0/go.mod h1:DnVip7tpJSsgZymwfT+m1tnEVy3ivAjSMXPx12YrMkU=
cel.dev/expr v0.25.3 h1:A2jO8jwOugrrovveCWfj0KEZOfqiLgAcwjpHPhzIGw0=
cel.dev/expr v0.25.3/go.mod h1:hrXvqGP6G6gyx8UAHSHJ5RGk//1Oj5nXQ2NI02Nrsg4=
dario.cat/mergo v1.0.2 h1:85+piFYR1tMbRrLcDwR18y4UKJ3aH1Tbzi24VRW1TK8=
dario.cat/mergo v1.0.2/go.mod h1:E/hbnu0NxMFBjpMIE34DRGLWqDy0g5FuKDhCb31ngxA=
github.com/AdaLogics/go-fuzz-headers v0.0.0-20240806141605-e8a1dd7889d6 h1:He8afgbRMd7mFxO99hRNu+6tazq8nFF9lIwo9JFroBk=
github.com/AdaLogics/go-fuzz-headers v0.0.0-20240806141605-e8a1dd7889d6/go.mod h1:8o94RPi1/7XTJvwPpRSzSUedZrtlirdB3r9Z20bi2f8=
github.com/Azure/go-ansiterm v0.0.0-20260917205352-e937bb47801a h1:4hxax9ktNjSDoFn1tZSeL6gXMTRMpO0FzjdKfT01jgY=
github.com/Azure/go-ansiterm v0.0.0-20260917205352-e937bb47801a/go.mod h1:3EWSSOZ50kb+arhww0qIaEXHToib/3FjBj9Jj5lcP5g=
github.com/BurntSushi/toml v1.6.0 h1:dRaEfpa2VI55EwlIW72hMRHdWouJeRF7TPYhI+AUQjk=
github.com/BurntSushi/toml v1.6.0/go.mod h1:ukJfTF/6rtPPRCnwkur4qwRxa8vTRFBF0uk2lLoLwho=
github.com/Microsoft/go-winio v0.6.2 h1:F2VQgta7ecxGYO8k3ZZz3RS8fVIXVxONVUPlNERoyfY=
github.com/Microsoft/go-winio v0.6.2/go.mod h1:yd8OoFMLzJbo9gZq8j5qaps8bJ9aShtEA8Ipt1oGCvU=
github.com/ProtonMail/go-crypto v1.5.1 h1:pTrLDQHyOT8y3DFYIpijgPBTw/7E2GLMimutvOlceuE=
github.com/ProtonMail/go-crypto v1.5.1/go.mod h1:/RaSu30DaKO4RY+XdV/ACcCcZkGr7AhUIduq5sjzzCo=
github.com/Microsoft/go-winio v0.6.4 h1:kJF2pamtYCWXf0xAT8sAkwKC0pMl2HYBrRPHAqyUc+g=
github.com/Microsoft/go-winio v0.6.4/go.mod h1:1xl5JXG4ze/m74GH6Dc4Mwxfc3VL5siRVQjIAlHH07U=
github.com/ProtonMail/go-crypto v1.5.2 h1:cucYnvqcY7UOXVD//mSyjeaPY0SSN3v5cDkYPxumINk=
github.com/ProtonMail/go-crypto v1.5.2/go.mod h1:/RaSu30DaKO4RY+XdV/ACcCcZkGr7AhUIduq5sjzzCo=
github.com/anmitsu/go-shlex v0.0.0-20200514113438-38f4b401e2be h1:9AeTilPcZAjCFIImctFaOjnTIavg87rW78vTPkQqLI8=
github.com/anmitsu/go-shlex v0.0.0-20200514113438-38f4b401e2be/go.mod h1:ySMOLuWl6zY27l47sB3qLNK6tF2fkHG55UZxx8oIVo4=
github.com/antithesishq/antithesis-sdk-go v0.8.0 h1:bVV5ZAuwaREn4klRSJusYaA+A7NXq8Qi9C/knCaQ2N4=
@@ -26,8 +20,6 @@ github.com/bufbuild/protocompile v0.14.1 h1:iA73zAf/fyljNjQKwYzUHD6AD4R8KMasmwa/
github.com/bufbuild/protocompile v0.14.1/go.mod h1:ppVdAIhbr2H8asPk6k4pY7t9zB1OU5DoEw9xY/FUi1c=
github.com/cedar-policy/cedar-go v1.8.0 h1:9gcU7EHXwHC2RMdpph68yTAkdB3behTTssC+kt4GoS8=
github.com/cedar-policy/cedar-go v1.8.0/go.mod h1:h5+3CVW1oI5LXVskJG+my9TFCYI5yjh/+Ul3EJie6MI=
github.com/cenkalti/backoff/v4 v4.3.0 h1:MyRJ/UdXutAwSAT+s3wNd7MfTIcy71VQueUuFK343L8=
github.com/cenkalti/backoff/v4 v4.3.0/go.mod h1:Y3VNntkOUPxTVeUxJ/G5vcM//AlwfmyYozVcomhLiZE=
github.com/cespare/xxhash/v2 v2.3.0 h1:UL815xU9SqsFlibzuggzjXhog7bL6oX9BbNZnL2UFvs=
github.com/cespare/xxhash/v2 v2.3.0/go.mod h1:VGX0DQ3Q6kWi7AoAeZDth3/j3BFtOZR5XLFGgcrjCOs=
github.com/cloudflare/circl v1.6.5 h1:O64F26HEqNhznd/hrC5KZXVKYuKM2rx4deZDTc4ihQA=
@@ -42,24 +34,18 @@ github.com/containerd/platforms v1.0.0-rc.5 h1:vXd569rDrz8LeMXzAnBsy6LADV5YtsD8o
github.com/containerd/platforms v1.0.0-rc.5/go.mod h1:lKlMXyLybmBedS/JJm11uDofzI8L2v0J2ZbYvNsbq1A=
github.com/coreos/go-oidc/v3 v3.21.0 h1:wZo4Q9Pum8dYEj0eMUPrqR+kvuGkeUplbLpNCkBqoWM=
github.com/coreos/go-oidc/v3 v3.21.0/go.mod h1:DYCf24+ncYi+XkIH97GY1+dqoRlbaSI26KVTCI9SrY4=
github.com/cpuguy83/dockercfg v0.3.2 h1:DlJTyZGBDlXqUZ2Dk2Q3xHs/FtnooJJVaad2S9GKorA=
github.com/cpuguy83/dockercfg v0.3.2/go.mod h1:sugsbF4//dDlL/i+S+rtpIWp+5h0BHJHfjj5/jFyUJc=
github.com/creack/pty v1.1.24 h1:bJrF4RRfyJnbTJqzRLHzcGaZK1NeM5kTC9jGgovnR1s=
github.com/creack/pty v1.1.24/go.mod h1:08sCNb52WyoAwi2QDyzUCTgcvVFhUzewun7wtTfvcwE=
github.com/davecgh/go-spew v1.1.0/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
github.com/davecgh/go-spew v1.1.1/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
github.com/distribution/reference v0.6.0 h1:0IXCQ5g4/QMHHkarYzh5l+u8T3t73zM5QvfrDyIgxBk=
github.com/distribution/reference v0.6.0/go.mod h1:BbU0aIcezP1/5jX/8MP0YiH4SdvB5Y4f/wlDRiLyi3E=
github.com/docker/go-connections v0.8.1 h1:JibmG5hULs5qXSr/cp/w3Pw5fZuStt4MOHMUExb29/M=
github.com/docker/go-connections v0.8.1/go.mod h1:no1qkHdjq7kLMGUXYAduOhYPSJxxvgWBh7ogVvptn3Q=
github.com/docker/go-connections v0.8.2 h1:9d9DEokx5W7qlr0shkFXIxEA7+TTD+llwzcykSlxMNc=
github.com/docker/go-connections v0.8.2/go.mod h1:no1qkHdjq7kLMGUXYAduOhYPSJxxvgWBh7ogVvptn3Q=
github.com/docker/go-units v0.5.0 h1:69rxXcBk27SvSaaxTtLh/8llcHD8vYHT7WSdRZ/jvr4=
github.com/docker/go-units v0.5.0/go.mod h1:fgPhTUdO+D/Jk86RDLlptpiXQzgHJF7gydDDbaIK4Dk=
github.com/docker/sandbox-kit-spec/v3 v3.0.0-m.6 h1:co7vn6GgImYEGO/KPlhPRJT9i5EPWgOKcB9q7bNJlFQ=
github.com/docker/sandbox-kit-spec/v3 v3.0.0-m.6/go.mod h1:MrWO4PN3/VYC5xF8eyI8U82Nv/agTHcN4IUzheqA90c=
github.com/docker/sandbox-kit-spec/v3 v3.0.0-m.8 h1:ay2XEj0yW3uzsNwN3KXS/c4nlneHONFKaHlrVhu2WCQ=
github.com/docker/sandbox-kit-spec/v3 v3.0.0-m.8/go.mod h1:4G6VgFAp9eaQkwy7DbZbOfWRD/Mjvc/TnEtQBRRMEsA=
github.com/dustin/go-humanize v1.1.0 h1:dbKTrvD0klcbBV/h4AWJdMuZogJACoMlvWIWZ5b2xWg=
github.com/dustin/go-humanize v1.1.0/go.mod h1:hc1CvRkJMsgxqjmjMQF3QNRAZBwY8AXBAzKYoSX9sFI=
github.com/ebitengine/purego v0.11.1 h1:2zpWRSQNVKN4eKsKO9eM1ILDgWfYMY9GwqRmK6XeQ/0=
github.com/ebitengine/purego v0.11.1/go.mod h1:DCHPP08djqhNSoTfImcnHYQRZmd0qhakvrozqaEYhGQ=
github.com/emirpasic/gods v1.18.1 h1:FXtiHYKDGKCW2KzwZKx0iC0PQmdlorYgdFG9jPXJ1Bc=
github.com/emirpasic/gods v1.18.1/go.mod h1:8tpGGwCnJ5H4r6BWwaV6OrWmMoPhUl5jm/FMNAnJvWQ=
github.com/fatih/color v1.13.0/go.mod h1:kLAiJbzzSOZDVNGyDpeOxJ47H46qBXwg5ILebYFFOfk=
@@ -67,16 +53,18 @@ github.com/fatih/color v1.19.0 h1:Zp3PiM21/9Ld6FzSKyL5c/BULoe/ONr9KlbYVOfG8+w=
github.com/fatih/color v1.19.0/go.mod h1:zNk67I0ZUT1bEGsSGyCZYZNrHuTkJJB+r6Q9VuMi0LE=
github.com/felixge/httpsnoop v1.1.0 h1:3YtUj32ZZkqZtt3sZZsClsymw/QDuVfpNhoA31zeORc=
github.com/felixge/httpsnoop v1.1.0/go.mod h1:Zqxgdd+1Rkcz8euOqdr7lqgCRJztwr5hp9vDSi5UZCE=
github.com/fxamacker/cbor/v2 v2.9.4 h1:xwjVlxEMR3S605oUlgBjKLTTeGFciYPGYCtF/35LKGo=
github.com/fxamacker/cbor/v2 v2.9.4/go.mod h1:vM4b+DJCtHn+zz7h3FFp/hDAI9WNWCsZj23V5ytsSxQ=
github.com/gliderlabs/ssh v0.3.8 h1:a4YXD1V7xMF9g5nTkdfnja3Sxy1PVDCj1Zg4Wb8vY6c=
github.com/gliderlabs/ssh v0.3.8/go.mod h1:xYoytBv1sV0aL3CavoDuJIQNURXkkfPA/wxQ1pL1fAU=
github.com/go-git/gcfg/v2 v2.0.2 h1:MY5SIIfTGGEMhdA7d7JePuVVxtKL7Hp+ApGDJAJ7dpo=
github.com/go-git/gcfg/v2 v2.0.2/go.mod h1:/lv2NsxvhepuMrldsFilrgct6pxzpGdSRC13ydTLSLs=
github.com/go-git/go-billy/v6 v6.0.0-alpha.2 h1:1Sv5WemXL8CxKrAx1gioJ+uHNb2bZJhiQLfwSZ4Et8c=
github.com/go-git/go-billy/v6 v6.0.0-alpha.2/go.mod h1:r/bsv9i/iDyyEU8/Z6mjC+YraOVwie1ddfUqBCElKXQ=
github.com/go-git/go-billy/v6 v6.0.0-beta.1 h1:OJ4r207zcEbeq/DJNajQUOMd+pph3r9A7hRq1Sjl1bA=
github.com/go-git/go-billy/v6 v6.0.0-beta.1/go.mod h1:QGOWN4CzmtRfpFeCw5paG3RJyKdTwsfGASkxBVkIaAU=
github.com/go-git/go-git-fixtures/v6 v6.0.0-alpha.1 h1:gmqi2jvsreu0s8JMLylYDFq4sbjHwwlhktMw0DUg3mA=
github.com/go-git/go-git-fixtures/v6 v6.0.0-alpha.1/go.mod h1:ECf1MqJlBdYpKggBrOXjo/0EnvRZx6D++I86UYjPgAQ=
github.com/go-git/go-git/v6 v6.0.0-alpha.5 h1:sE+OlkHgYWNMVmN1s9sR7uyFgsWLtxcNWse/vBYKxRE=
github.com/go-git/go-git/v6 v6.0.0-alpha.5/go.mod h1:3IjhiZnM+uBmUrOGSeqrJpsmi4Vd0H2NZO/uK2a7d0s=
github.com/go-git/go-git/v6 v6.0.0-beta.1 h1:1t+vJYGDBN6f9GRWskgWx5UjXPFUa7wfffZZZ2C5GtA=
github.com/go-git/go-git/v6 v6.0.0-beta.1/go.mod h1:ugDlr38Wwc+fOw85J+Oluh5TNXQtJgYMcriOljg3l4s=
github.com/go-jose/go-jose/v4 v4.1.5 h1:RjgjO2LOtWOJKUC5wpwY9LR3B3vwVAz6JS2YHfYU6eA=
github.com/go-jose/go-jose/v4 v4.1.5/go.mod h1:x4oUasVrzR7071A4TnHLGSPpNOm2a21K9Kf04k1rs08=
github.com/go-json-experiment/json v0.0.0-20260820222146-c27c302e5fc3 h1:UADEEmDKgfXbtnGJZ97beY5XLo9ZechG1nlU4KnRrkE=
@@ -86,11 +74,14 @@ github.com/go-logr/logr v1.4.4 h1:tG4xh9yMsRCAiodLVTxyrkzSZ9+o0L1Kg/+cPVcbP/8=
github.com/go-logr/logr v1.4.4/go.mod h1:9T104GzyrTigFIr8wt5mBrctHMim0Nb2HLGrmQ40KvY=
github.com/go-logr/stdr v1.2.2 h1:hSWxHoqTgW2S2qGc0LTAI563KZ5YKYRhT3MFKZMbjag=
github.com/go-logr/stdr v1.2.2/go.mod h1:mMo/vtBO5dYbehREoey6XUKy/eSumjCCveDpRre4VKE=
github.com/go-ole/go-ole v1.2.6/go.mod h1:pprOEPIfldk/42T2oK7lQ4v4JSDwmV0As9GaiUsvbm0=
github.com/go-ole/go-ole v1.3.0 h1:Dt6ye7+vXGIKZ7Xtk4s6/xVdGDQynvom7xCFEdWr6uE=
github.com/go-ole/go-ole v1.3.0/go.mod h1:5LS6F96DhAwUc7C+1HLexzMXY1xGRSryjyPPKW6zv78=
github.com/go-telegram/bot v1.27.0 h1:PG9yBDzM8O26oUerE9SZwWlHon4RpH8WcNslZ7CEZmE=
github.com/go-telegram/bot v1.27.0/go.mod h1:i2TRs7fXWIeaceF3z7KzsMt/he0TwkVC680mvdTFYeM=
github.com/go-viper/mapstructure/v2 v2.5.0 h1:vM5IJoUAy3d7zRSVtIwQgBj7BiWtMPfmPEgAXnvj1Ro=
github.com/go-viper/mapstructure/v2 v2.5.0/go.mod h1:oJDH3BJKyqBA2TXFhDsKDGDTlndYOZ6rGS0BRZIxGhM=
github.com/go-webauthn/webauthn v0.18.2 h1:0BeftmEHU7i3Dv0VFwBtidy/ba37Vcdjvqst9EYu8Sk=
github.com/go-webauthn/webauthn v0.18.2/go.mod h1:hEXaOuLxvZ3zG9miZe3ehlyeVso9AtklXG+kTn36k+A=
github.com/go-webauthn/x v0.3.1 h1:1ff37z3XfmTTomkhlURgGizLIDyOvPgTt2t9nlzKLRo=
github.com/go-webauthn/x v0.3.1/go.mod h1:ZInxAynYXfBPvvm5gzKZ7geBlL23K71xASMgohHl/Rg=
github.com/golang-jwt/jwt/v5 v5.3.1 h1:kYf81DTWFe7t+1VvL7eS+jKFVWaUnK9cB1qbwn63YCY=
github.com/golang-jwt/jwt/v5 v5.3.1/go.mod h1:fxCRLWMO43lRc8nhHWY6LGqRcf+1gQWArsqaEUEa5bE=
github.com/golang/protobuf v1.5.4 h1:i7eJL8qZTpSEXOPTxNKhASYpMn+8e5Q6AdndVa1dWek=
@@ -99,6 +90,8 @@ github.com/google/go-cmp v0.7.0 h1:wk8382ETsv4JYUZwIsn6YpYiWiBsYLSJiTsyBybVuN8=
github.com/google/go-cmp v0.7.0/go.mod h1:pXiqmnSA92OHEEa9HXL2W4E7lf9JzCmGVUdgjX3N/iU=
github.com/google/go-tpm v0.9.8 h1:slArAR9Ft+1ybZu0lBwpSmpwhRXaa85hWtMinMyRAWo=
github.com/google/go-tpm v0.9.8/go.mod h1:h9jEsEECg7gtLis0upRBQU+GhYVH6jMjrFxI8u6bVUY=
github.com/google/go-tpm-tools v0.3.13-0.20230620182252-4639ecce2aba h1:qJEJcuLzH5KDR0gKc0zcktin6KSAwL7+jWKBYceddTc=
github.com/google/go-tpm-tools v0.3.13-0.20230620182252-4639ecce2aba/go.mod h1:EFYHy8/1y2KfgTAsx7Luu7NGhoxtuVHnNo8jE7FikKc=
github.com/google/jsonschema-go v0.4.3 h1:/DBOLZTfDow7pe2GmaJNhltueGTtDKICi8V8p+DQPd0=
github.com/google/jsonschema-go v0.4.3/go.mod h1:r5quNTdLOYEz95Ru18zA0ydNbBuYoo9tgaYcxEYhJVE=
github.com/google/uuid v1.6.0 h1:NIvaJDMOsjHA8n1jAhLSgzrAzy1Hgr+hNrb57e+94F0=
@@ -115,16 +108,14 @@ github.com/jackc/pgservicefile v0.0.0-20240606120523-5a60cdf6a761 h1:iCEnooe7Ulw
github.com/jackc/pgservicefile v0.0.0-20240606120523-5a60cdf6a761/go.mod h1:5TJZWKEWniPve33vlWYSoGYefn3gLQRzjfDlhSJ9ZKM=
github.com/jackc/pgx/v5 v5.11.0 h1:IzBBtyK9AHqf98cctWFifYSci2hgQR/cd56wB4p+ogg=
github.com/jackc/pgx/v5 v5.11.0/go.mod h1:mal1tBGAFfLHvZzaYh77YS/eC6IX9OWbRV1QIIM0Jn4=
github.com/jackc/puddle/v2 v2.2.2 h1:PR8nw+E/1w0GLuRFSmiioY6UooMp6KJv0/61nB7icHo=
github.com/jackc/puddle/v2 v2.2.2/go.mod h1:vriiEXHvEE654aYKXXjOvZM39qJ0q+azkZFrfEOc3H4=
github.com/jackc/puddle/v2 v2.2.3 h1:f8HoREISb3l5eO9coPgK7U3hKLhWGokOFVcThl5jrao=
github.com/jackc/puddle/v2 v2.2.3/go.mod h1:vriiEXHvEE654aYKXXjOvZM39qJ0q+azkZFrfEOc3H4=
github.com/jhump/protoreflect v1.17.0 h1:qOEr613fac2lOuTgWN4tPAtLL7fUSbuJL5X5XumQh94=
github.com/jhump/protoreflect v1.17.0/go.mod h1:h9+vUUL38jiBzck8ck+6G/aeMX8Z4QUY/NiJPwPNi+8=
github.com/kevinburke/ssh_config v1.6.0 h1:J1FBfmuVosPHf5GRdltRLhPJtJpTlMdKTBjRgTaQBFY=
github.com/kevinburke/ssh_config v1.6.0/go.mod h1:q2RIzfka+BXARoNexmF9gkxEX7DmvbW9P4hIVx2Kg4M=
github.com/klauspost/compress v1.20.0 h1:a3C1ke2ohxFymNlb2HWAHjDeKCI90scRskErZkR0ezA=
github.com/klauspost/compress v1.20.0/go.mod h1:LUdAzn7YLVvxLpc7y3V1m40wESHTgc1422pwwBSKYuI=
github.com/klauspost/cpuid/v2 v2.4.0 h1:S6Hrbc7+ywsr0r+RLapfGBHfyefhCTwEh3A0tV913Dw=
github.com/klauspost/cpuid/v2 v2.4.0/go.mod h1:19jmZ9mjzoF//ddRSUsv0zfBTJWh3QJh9FNxZTMrGxU=
github.com/klauspost/compress v1.20.3 h1:zWNoFqq/lSuYQRim3dUXlXeA0IqwuZGicdG5ZTfY5j4=
github.com/klauspost/compress v1.20.3/go.mod h1:LUdAzn7YLVvxLpc7y3V1m40wESHTgc1422pwwBSKYuI=
github.com/kr/pretty v0.1.0/go.mod h1:dAy3ld7l9f0ibDNOQOHHMYYIIbhfbHSm3C4ZsoJORNo=
github.com/kr/pretty v0.3.1 h1:flRD4NNwYAUpkphVc1HcthR4KEIFJ65n8Mw5qdRn3LE=
github.com/kr/pretty v0.3.1/go.mod h1:hoEshYVHaxMs3cyo3Yncou5ZscifuDolrwPKZanG3xk=
@@ -132,116 +123,82 @@ github.com/kr/pty v1.1.1/go.mod h1:pFQYn66WHrOpPYNljwOMqo10TkYh1fy3cYio2l3bCsQ=
github.com/kr/text v0.1.0/go.mod h1:4Jbv+DJW3UT/LiOwJeYQe1efqtUx/iVham/4vfdArNI=
github.com/kr/text v0.2.0 h1:5Nx0Ya0ZqY2ygV366QzturHI13Jq95ApcVaJBhpS+AY=
github.com/kr/text v0.2.0/go.mod h1:eLer722TekiGuMkidMxC/pM04lWEeraHUUmBw8l2grE=
github.com/lib/pq v1.10.9 h1:YXG7RB+JIjhP29X+OtkiDnYaXQwpS4JEWq7dtCCRUEw=
github.com/lib/pq v1.10.9/go.mod h1:AlVN5x4E4T544tWzH6hKfbfQvm3HdbOxrmggDNAPY9o=
github.com/lufia/plan9stats v0.0.0-20260802145828-341c2f0c90b5 h1:eveIIGn4BGM3qknO74omf6HYr30/exH+eVUTuAgwjZ0=
github.com/lufia/plan9stats v0.0.0-20260802145828-341c2f0c90b5/go.mod h1:autxFIvghDt3jPTLoqZ9OZ7s9qTGNAWmYCjVFWPX/zg=
github.com/magiconair/properties v1.18.12 h1:sT9zQpvTB3B4gzrX0tmZNTEaGyg8Zw55MFYRE32Mr9I=
github.com/magiconair/properties v1.18.12/go.mod h1:Dhd985XPs7jluiymwWYZ0G4Z61jb3vdS329zhj2hYo0=
github.com/mattn/go-colorable v0.1.9/go.mod h1:u6P/XSegPjTcexA+o6vUJrdnUu04hMope9wVRipJSqc=
github.com/mattn/go-colorable v0.1.12/go.mod h1:u5H1YNBxpqRaxsYJYSkiCWKzEfiAb1Gb520KVy5xxl4=
github.com/mattn/go-colorable v0.1.14 h1:9A9LHSqF/7dyVVX6g0U9cwm9pG3kP9gSzcuIPHPsaIE=
github.com/mattn/go-colorable v0.1.14/go.mod h1:6LmQG8QLFO4G5z1gPvYEzlUgJ2wF+stgPZH1UqBm1s8=
github.com/mattn/go-colorable v0.1.16 h1:+2YjeAe7xP1OQbZW1jERPWWCDLlNqFNCvjiqGiAt84A=
github.com/mattn/go-colorable v0.1.16/go.mod h1:6LmQG8QLFO4G5z1gPvYEzlUgJ2wF+stgPZH1UqBm1s8=
github.com/mattn/go-isatty v0.0.12/go.mod h1:cbi8OIDigv2wuxKPP5vlRcQ1OAZbq2CE4Kysco4FUpU=
github.com/mattn/go-isatty v0.0.14/go.mod h1:7GGIvUiUoEMVVmxf/4nioHXj79iQHKdU27kJ6hsGG94=
github.com/mattn/go-isatty v0.0.24 h1:tGZZoVgT/KiqK1c8ocVLeDS8BSWMRd47J3Lbz7vsReI=
github.com/mattn/go-isatty v0.0.24/go.mod h1:nMCL3Zebbrt45jsMDgnfIwz6ydEQApk5oEI3HqDio6A=
github.com/mdelapenya/tlscert v0.2.0 h1:7H81W6Z/4weDvZBNOfQte5GpIMo0lGYEeWbkGp5LJHI=
github.com/mdelapenya/tlscert v0.2.0/go.mod h1:O4njj3ELLnJjGdkN7M/vIVCpZ+Cf0L6muqOG4tLSl8o=
github.com/mfridman/interpolate v0.0.2 h1:pnuTK7MQIxxFz1Gr+rjSIx9u7qVjf5VOoM/u6BbAxPY=
github.com/mfridman/interpolate v0.0.2/go.mod h1:p+7uk6oE07mpE/Ik1b8EckO0O4ZXiGAfshKBWLUM9Xg=
github.com/minio/highwayhash v1.0.4 h1:asJizugGgchQod2ja9NJlGOWq4s7KsAWr5XUc9Clgl4=
github.com/minio/highwayhash v1.0.4/go.mod h1:GGYsuwP/fPD6Y9hMiXuapVvlIUEhFhMTh0rxU3ik1LQ=
github.com/moby/docker-image-spec v1.3.1 h1:jMKff3w6PgbfSa69GfNg+zN/XLhfXJGnEx3Nl2EsFP0=
github.com/moby/docker-image-spec v1.3.1/go.mod h1:eKmb5VW8vQEh/BAr2yvVNvuiJuY6UIocYsFu/DxxRpo=
github.com/moby/go-archive v0.3.3 h1:OxxR9paxsluYi+zDUEXTTaIxtkK3viymW+Ka7vRhhME=
github.com/moby/go-archive v0.3.3/go.mod h1:Npdv43fFqlhZW7Xo8fbm3ZMYFvAGNviUPqX21VERbcE=
github.com/moby/moby/api v1.56.0 h1:GQzua3NA599ASSIICx0iFgiJeO9YkdDARvQsm23ZZuQ=
github.com/moby/moby/api v1.56.0/go.mod h1:sZ+THbVWkjOmBPPfbnzdD/G1LuIexWhqlSHHPTDQ1Uk=
github.com/moby/moby/client v0.6.0 h1:AJjEB21QPbXSXjDsZorFBoDZPhMrfbpaPLgSMAW9Bgs=
github.com/moby/moby/client v0.6.0/go.mod h1:OCo00wNRyA3m4lmJ228W3JbyCN4ZNNYjpOXiJydBdcQ=
github.com/moby/patternmatcher v0.6.1 h1:qlhtafmr6kgMIJjKJMDmMWq7WLkKIo23hsrpR3x084U=
github.com/moby/patternmatcher v0.6.1/go.mod h1:hDPoyOpDY7OrrMDLaYoY3hf52gNCR/YOUYxkhApJIxc=
github.com/moby/sys/mount v0.3.5 h1:eS3fsZTjHaBihwjp4/+5Z3jxqLXYsbwxqpVSfFv3M00=
github.com/moby/sys/mount v0.3.5/go.mod h1:WUQDO+/uCiCIkIztx8SrwIDVn2dtMFRBebRhpDFT71M=
github.com/moby/sys/mountinfo v0.7.2 h1:1shs6aH5s4o5H2zQLn796ADW1wMrIwHsyJ2v9KouLrg=
github.com/moby/sys/mountinfo v0.7.2/go.mod h1:1YOa8w8Ih7uW0wALDUgT1dTTSBrZ+HiBLGws92L2RU4=
github.com/moby/sys/sequential v0.7.0 h1:ASQNGNROJSuOO6LL6bPHbKvuZu6NU8P4ldPWk31zj/8=
github.com/moby/sys/sequential v0.7.0/go.mod h1:NfSTAp6V3fw4tmkD62PEcOKeZKquXT8VKCkf7aVR79o=
github.com/moby/sys/user v0.4.1 h1:RgjRlaDKi/Xmyrz4t8lyzXT6v2ooFeO/7xtchmhVWE0=
github.com/moby/sys/user v0.4.1/go.mod h1:E9QsW5WRe1kUAf7kW8hXKwu1uhsZEAdPLYHYSDudF4Y=
github.com/moby/sys/userns v0.2.1 h1:4OvdM7BcPkASbuouHsbW3aeMJSFlYDldBRnXVZhaRk8=
github.com/moby/sys/userns v0.2.1/go.mod h1:IHUYgu/kao6N8YZlp9Cf444ySSvCmDlmzUcYfDHOl28=
github.com/moby/term v0.5.2 h1:6qk3FJAFDs6i/q3W/pQ97SX192qKfZgGjCQqfCJkgzQ=
github.com/moby/term v0.5.2/go.mod h1:d3djjFCrjnB+fl8NJux+EJzu0msscUP+f8it8hPkFLc=
github.com/moby/moby/api v1.56.1 h1:PpWkvVPB7Fr/No8w+TfyJ/I6rWZ2YPZhT3JorIg+06c=
github.com/moby/moby/api v1.56.1/go.mod h1:sZ+THbVWkjOmBPPfbnzdD/G1LuIexWhqlSHHPTDQ1Uk=
github.com/moby/moby/client v0.6.2 h1:ekSPlMccrZGcw6Scb8T9D39qE0mB+XupU/yS0zNw9N4=
github.com/moby/moby/client v0.6.2/go.mod h1:DGcnO7qmPJrqobx/ggQgQ/AUk+2wgVdQv7uxB0/mEsc=
github.com/modelcontextprotocol/go-sdk v1.8.0 h1:KIvahhYqwtbeniWVPs3TcXEA7b8jEtwfBpOTAI+Urx4=
github.com/modelcontextprotocol/go-sdk v1.8.0/go.mod h1:dL7u98E/zjJTGzEq+j30jQ8K2k1mb6LeAH4inEcSGts=
github.com/nats-io/jwt/v2 v2.8.2 h1:XXRgB60MSTnqsRwejQurVDs/hcv2dkt+86GjI+I/bMc=
github.com/nats-io/jwt/v2 v2.8.2/go.mod h1:Ag/56sq9OblL4JgdYufDd16Egb17Kr/8WwwuO/forVc=
github.com/nats-io/nats-server/v2 v2.15.0 h1:M99yf0y05rTr46/qc/Is6ZAowI58Ryp2SjufLCUeVJc=
github.com/nats-io/nats-server/v2 v2.15.0/go.mod h1:5qLF4CDGzZVFt//3fUrY1ePpwbi05r7QHPNroSUtolk=
github.com/nats-io/nats-server/v2 v2.15.1 h1:fK3AeYFIji1Smng60jB0hVoAPZezA/UzsaI3BqXFNEI=
github.com/nats-io/nats-server/v2 v2.15.1/go.mod h1:yeuYRtKVKIFtmMz2/YPk+6hvmNBnk+WwTUzErBdH6IE=
github.com/nats-io/nats.go v1.54.0 h1:vsXoOxjHp/GmPUN+EcI7uOf/uB+iAP+kEsAFNQN0yzA=
github.com/nats-io/nats.go v1.54.0/go.mod h1:y+DZoD1oBOYfZTU681eTUiUjI0vbqYGixNVFHcjHJ0k=
github.com/nats-io/nkeys v0.4.16 h1:rd5oAuLOb8mnAycB0xleuEBNS1pVVnN0fv/FF34Eypg=
github.com/nats-io/nkeys v0.4.16/go.mod h1:llLgWoI0o4z/Q57q2R1kHfmocyhGV6VG/U18Glg1Afs=
github.com/nats-io/nuid v1.0.1 h1:5iA8DT8V7q8WK2EScv2padNa/rTESc1KdnPw4TC2paw=
github.com/nats-io/nuid v1.0.1/go.mod h1:19wcPz3Ph3q0Jbyiqsd0kePYG7A95tJPxeL+1OSON2c=
github.com/ncruces/go-strftime v1.0.0 h1:HMFp8mLCTPp341M/ZnA4qaf7ZlsbTc+miZjCLOFAw7w=
github.com/ncruces/go-strftime v1.0.0/go.mod h1:Fwc5htZGVVkseilnfgOVb9mKy6w1naJmn9CehxcKcls=
github.com/oklog/run v1.1.0 h1:GEenZ1cK0+q0+wsJew9qUg/DyD8k3JzYsZAi5gYi2mA=
github.com/oklog/run v1.1.0/go.mod h1:sVPdnTZT1zYwAJeCMu2Th4T21pA3FPOQRfWjQlk7DVU=
github.com/ncruces/go-strftime v1.1.0 h1:vtPmPWMkb539RGV9WpWT9EwJGvrOCKvMNI4obWL47Hk=
github.com/ncruces/go-strftime v1.1.0/go.mod h1:yVaq7iwCN09vHRrKBrrflXT0M9tTMG2KmZ9Fj7xeG/g=
github.com/oklog/run v1.2.0 h1:O8x3yXwah4A73hJdlrwo/2X6J62gE5qTMusH0dvz60E=
github.com/oklog/run v1.2.0/go.mod h1:mgDbKRSwPhJfesJ4PntqFUbKQRZ50NgmZTSPlFA0YFk=
github.com/opencontainers/go-digest v1.0.0 h1:apOUWs51W5PlhuyGyz9FCeeBIOUDA/6nW8Oi/yOhh5U=
github.com/opencontainers/go-digest v1.0.0/go.mod h1:0JzlMkj0TRzQZfJkVvzbP0HBR3IKzErnv2BNG4W4MAM=
github.com/opencontainers/image-spec v1.1.1 h1:y0fUlFfIZhPF1W537XOLg0/fcx6zcHCJwooC2xJA040=
github.com/opencontainers/image-spec v1.1.1/go.mod h1:qpqAh3Dmcf36wStyyWU+kCeDgrGnAve2nCC8+7h8Q0M=
github.com/pjbgf/sha1cd v0.6.0 h1:3WJ8Wz8gvDz29quX1OcEmkAlUg9diU4GxJHqs0/XiwU=
github.com/pjbgf/sha1cd v0.6.0/go.mod h1:lhpGlyHLpQZoxMv8HcgXvZEhcGs0PG/vsZnEJ7H0iCM=
github.com/philhofer/fwd v1.2.0 h1:e6DnBTl7vGY+Gz322/ASL4Gyp1FspeMvx1RNDoToZuM=
github.com/philhofer/fwd v1.2.0/go.mod h1:RqIHx9QI14HlwKwm98g9Re5prTQ6LdeRQn+gXJFxsJM=
github.com/pjbgf/sha1cd v0.7.0 h1:ZRNPKHj+gfkLBf0KJv/p1Hmz+7bqCT5o311YX0nq+DA=
github.com/pjbgf/sha1cd v0.7.0/go.mod h1:pKR5Li+qTCo+ebqITZfwPlJGoCFCvSO00qxjbNtTUFQ=
github.com/pkg/browser v0.0.0-20240102092130-5ac0b6a4141c h1:+mdjkGKdHQG3305AYmdv1U2eRNDiU2ErMBj1gwrq8eQ=
github.com/pkg/browser v0.0.0-20240102092130-5ac0b6a4141c/go.mod h1:7rwL4CYBLnjLxUqIJNnCWiEdr3bn6IUYi15bNlnbCCU=
github.com/pmezard/go-difflib v1.0.0/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4=
github.com/power-devops/perfstat v0.0.0-20260916203055-22a1a467d9f0 h1:XA01Vk/wv9YikCi1V51yRzIHPMT5of9+cMpZoZvDn/M=
github.com/power-devops/perfstat v0.0.0-20260916203055-22a1a467d9f0/go.mod h1:OmDBASR4679mdNQnz2pUhc2G8CO2JrUAVFDRBDP/hJE=
github.com/pressly/goose/v3 v3.28.0 h1:D2M+iL31GmpZxSHOhX8mqyqAT3CXnokUmm0eKoSP+Vc=
github.com/pressly/goose/v3 v3.28.0/go.mod h1:v26MOuB8bL3kzzrt3Vqhb3R0PRVsl8hFQKdrht/L6Rk=
github.com/pressly/goose/v3 v3.29.0 h1:dtEnBCT7CqkHIUCdgrXsrmeisaN8gFkRU4Y5OEwif4o=
github.com/pressly/goose/v3 v3.29.0/go.mod h1:plA4aP4+u+Q3FMTgbFbhF8TuAWpAdfWpoFGkJsIFWh0=
github.com/remyoudompheng/bigfft v0.0.0-20230129092748-24d4a6f8daec h1:W09IVJc94icq4NjY3clb7Lk8O1qJ8BdBEF8z0ibU0rE=
github.com/remyoudompheng/bigfft v0.0.0-20230129092748-24d4a6f8daec/go.mod h1:qqbHyh8v60DhA7CoWK5oRCqLrMHRGoxYCSS9EjAz6Eo=
github.com/rogpeppe/go-internal v1.14.1 h1:UQB4HGPB6osV0SQTLymcB4TgvyWu6ZyliaW0tI/otEQ=
github.com/rogpeppe/go-internal v1.14.1/go.mod h1:MaRKkUm5W0goXpeCfT7UZI6fk/L7L7so1lCWt35ZSgc=
github.com/samcharles93/ai-sdk v0.1.42 h1:YV4O6oV0OLIqDAnEWAotPyw19K1VH2YoAT0He7aW+Lg=
github.com/samcharles93/ai-sdk v0.1.42/go.mod h1:u4zrQ9/Q58L5mNEut4bNQ1ks+RCKq71LN5EDz2N/0Zk=
github.com/samcharles93/ai-sdk v0.3.0 h1:ZZX38sT00oJkbry/cgjIaFnx5IHGHorfStPrVcGCjVU=
github.com/samcharles93/ai-sdk v0.3.0/go.mod h1:u4zrQ9/Q58L5mNEut4bNQ1ks+RCKq71LN5EDz2N/0Zk=
github.com/segmentio/asm v1.2.1 h1:DTNbBqs57ioxAD4PrArqftgypG4/qNpXoJx8TVXxPR0=
github.com/segmentio/asm v1.2.1/go.mod h1:BqMnlJP91P8d+4ibuonYZw9mfnzI9HfxselHZr5aAcs=
github.com/segmentio/encoding v0.5.4 h1:OW1VRern8Nw6ITAtwSZ7Idrl3MXCFwXHPgqESYfvNt0=
github.com/segmentio/encoding v0.5.4/go.mod h1:HS1ZKa3kSN32ZHVZ7ZLPLXWvOVIiZtyJnO1gPH1sKt0=
github.com/sergi/go-diff v1.4.0 h1:n/SP9D5ad1fORl+llWyN+D6qoUETXNZARKjyY2/KVCw=
github.com/sergi/go-diff v1.4.0/go.mod h1:A0bzQcvG0E7Rwjx0REVgAGH58e96+X0MeOfepqsbeW4=
github.com/sethvargo/go-retry v0.4.0 h1:9qy1OoIAxBL+gBYnkTnTnWle5wlfsXQlwRzIbbpdqPw=
github.com/sethvargo/go-retry v0.4.0/go.mod h1:tvsjdKG6xfiCx4LSiUZ06kcv38xvdVQwv8R6/VnnVWg=
github.com/shirou/gopsutil/v4 v4.26.8 h1:YQMTF/1J50B5+Y0vlo1eDRf5DoR7Gk69hY+8wjYkQeo=
github.com/shirou/gopsutil/v4 v4.26.8/go.mod h1:5O9FjBiXoTDFatIWjZZosqj4pV0DRtLx598xGbBehzM=
github.com/sethvargo/go-retry v0.5.0 h1:JYfFKmVJ8ER06GReN1efzc5cm52BrVsFVzmcHiKJwfY=
github.com/sethvargo/go-retry v0.5.0/go.mod h1:tvsjdKG6xfiCx4LSiUZ06kcv38xvdVQwv8R6/VnnVWg=
github.com/sirupsen/logrus v1.10.2 h1:G2SED73/qrAu6YwbdxOD6peLkCBI3z7L+ykJFTXJBBo=
github.com/sirupsen/logrus v1.10.2/go.mod h1:SLEg8TqYulVKKfIGHldVp2K2aYz2DKSVBq4g/H5bR7Q=
github.com/stretchr/objx v0.1.0/go.mod h1:HFkY916IF+rwdDfMAkV7OtwuqBVzrE8GR6GFx+wExME=
github.com/stretchr/objx v0.5.3 h1:jmXUvGomnU1o3W/V5h2VEradbpJDwGrzugQQvL0POH4=
github.com/stretchr/objx v0.5.3/go.mod h1:rDQraq+vQZU7Fde9LOZLr8Tax6zZvy4kuNKF+QYS+U0=
github.com/stretchr/testify v1.3.0/go.mod h1:M5WIy9Dh21IEIfnGCwXGc5bZfKNJtfHm1UVUgZn+9EI=
github.com/stretchr/testify v1.4.0/go.mod h1:j7eGeouHqKxXV5pUuKE4zz7dFj8WfuZ+81PSLYec5m4=
github.com/stretchr/testify v1.7.0/go.mod h1:6Fq8oRcR53rry900zMqJjRRixrwX3KX962/h/Wwjteg=
github.com/stretchr/testify v1.7.2/go.mod h1:R6va5+xMeoiuVRoj+gSkQ7d3FALtqAAGI1FQKckRals=
github.com/stretchr/testify v1.12.1 h1:EuwCh5fleGS7H32xRwO3wRGT7DxrDhLAT6FF8MpWDWE=
github.com/stretchr/testify v1.12.1/go.mod h1:MDEgiDPPsNp5cuIrHPPCyornHKgEVbtFUmoNlxoYthg=
github.com/testcontainers/testcontainers-go v0.44.0 h1:/Fwh6HY1mIikhnm9e7HwoxGycx0lzRAE0f5VQpjFxzI=
github.com/testcontainers/testcontainers-go v0.44.0/go.mod h1:IcnwQrYTO86xHXu5bvMaBH7ATlbS3Qn1M1QWW3c66rE=
github.com/testcontainers/testcontainers-go/modules/postgres v0.44.0 h1:8fdv/9y3JMxjQ+ULAcOG8RtgeNu5t9XF9LolSXDuTwM=
github.com/testcontainers/testcontainers-go/modules/postgres v0.44.0/go.mod h1:CFr2LncGYokw+OKjXcr8ARCKG1SaC2UEnGxFBovE86g=
github.com/tklauser/go-sysconf v0.4.0 h1:7H0uAN+7RkwWRaxhYXDLqa5V3LPrJeV8wmD9dRUgPQU=
github.com/tklauser/go-sysconf v0.4.0/go.mod h1:8mTNWyog7H+MpKijp4VmKJAd2bbYQ2zuUwkYRbUArPI=
github.com/tklauser/numcpus v0.12.0 h1:NR85qdvHA9pFse3x3weVZ0r0ST8R6l5RHbZrlRaqob4=
github.com/tklauser/numcpus v0.12.0/go.mod h1:ABHeXzJnr/qqwguhClkZKT1/8VABcYrsyUiUGobwWJg=
github.com/tinylib/msgp v1.6.4 h1:mOwYbyYDLPj35mkA2BjjYejgJk9BuHxDdvRnb6v2ZcQ=
github.com/tinylib/msgp v1.6.4/go.mod h1:RSp0LW9oSxFut3KzESt5Voq4GVWyS+PSulT77roAqEA=
github.com/x448/float16 v0.8.4 h1:qLwI1I70+NjRFUR3zs1JPUCgaCXSh3SW62uAKT1mSBM=
github.com/x448/float16 v0.8.4/go.mod h1:14CWIYCyZA/cWjXOioeEpHeN/83MdbZDRQHoFcYsOfg=
github.com/yosida95/uritemplate/v3 v3.0.2 h1:Ed3Oyj9yrmi9087+NczuL5BwkIc4wvTb5zIM+UJPGz4=
github.com/yosida95/uritemplate/v3 v3.0.2/go.mod h1:ILOh0sOhIJR3+L/8afwt/kE++YT040gmv5BQTMR2HP4=
github.com/yusufpapurcu/wmi v1.2.4 h1:zFUKzehAFReQwLys1b/iSMl+JQGSCSjtVqQn9bBrPo0=
github.com/yusufpapurcu/wmi v1.2.4/go.mod h1:SBZ9tNy3G9/m5Oi98Zks0QjeHVDvuK0qfxQmPyzfmi0=
go.lsp.dev/jsonrpc2 v1.0.1 h1:JZAchpm6RcOuwEOnZ7kF3kA1EpEjdJZ/6rMSfOF3+F4=
go.lsp.dev/jsonrpc2 v1.0.1/go.mod h1:WN+TesqrPumBmUdsuWyn9DHq3zl4i85L/qG0NLdj8qg=
go.lsp.dev/protocol v1.0.1 h1:kLWUIk1v/deXxo1D0VO4avjZnfeuZ82mouJgJd/NbdU=
@@ -250,62 +207,63 @@ go.lsp.dev/uri v1.0.1 h1:ywxJN1UXoQ+v0I9lkoQNMmsm7lub5G9ZG7KXT8YTx30=
go.lsp.dev/uri v1.0.1/go.mod h1:06a0ghafs4PuqkLsJGUmlgPHw6+kH79edRXjOT/rA8s=
go.opentelemetry.io/auto/sdk v1.2.1 h1:jXsnJ4Lmnqd11kwkBV2LgLoFMZKizbCi5fNZ/ipaZ64=
go.opentelemetry.io/auto/sdk v1.2.1/go.mod h1:KRTj+aOaElaLi+wW1kO/DZRXwkF4C5xPbEe3ZiIhN7Y=
go.opentelemetry.io/contrib/instrumentation/net/http/otelhttp v0.71.0 h1:3g7B90UzBltIDKq1/5mrTGxTnOFDV0ICOhLoxiZ8jlg=
go.opentelemetry.io/contrib/instrumentation/net/http/otelhttp v0.71.0/go.mod h1:Ef8SuTh59BT7+ofpDxN9z+yOlc4t2GjLmKDgYNJL/NU=
go.opentelemetry.io/otel v1.46.0 h1:FHt5/CDyVxi/8IM1CH7VE/rRgq3kLHa2mSTVMO8AWyc=
go.opentelemetry.io/otel v1.46.0/go.mod h1:Gj3SEScelsNC45tp4nSxRYlS+f5iez7W8XPMCt905kE=
go.opentelemetry.io/otel/metric v1.46.0 h1:yBnkXvgV7AXFILZc5K6IZe/CBFF3OS7BJ8ov6/lj0K8=
go.opentelemetry.io/otel/metric v1.46.0/go.mod h1:iPmdWqifKUdzziPkvvzIJXITl56fQx2mGM/DHLB3/2o=
go.opentelemetry.io/otel/sdk v1.46.0 h1:h5CNQQjEbuQXY/JfZtgt3i7HVFV3aHPO2OAwO2eTYPI=
go.opentelemetry.io/otel/sdk v1.46.0/go.mod h1:GAERFXFt5SYCEB+YiKUbMBeza6UaDH7GmGOZEfh2gSM=
go.opentelemetry.io/otel/sdk/metric v1.46.0 h1:0piZ26EG4RBfebb2jhDH6ERCYHoVWduc3kLgPCwSnSE=
go.opentelemetry.io/otel/sdk/metric v1.46.0/go.mod h1:I1PbKrdVc8Qu8HYVDNtqVIwLwjNrhsV/uFuxfwg8mO4=
go.opentelemetry.io/otel/trace v1.46.0 h1:OULy7ccdJnZtJ0UDYFOIGaCmiWzJ8Vi2G/Rsu60qs1c=
go.opentelemetry.io/otel/trace v1.46.0/go.mod h1:J7GAXweO77XSFkB/rmAqk9D6ihszhFjLU+d9WuUxDLI=
go.opentelemetry.io/contrib/instrumentation/net/http/otelhttp v0.72.0 h1:LxwW/9ctSCv+QkE/cLR7M91ZIkXNMqJtEMi1vCw9U8s=
go.opentelemetry.io/contrib/instrumentation/net/http/otelhttp v0.72.0/go.mod h1:tOsftB4SslBwwErVEPaenU2RpThXWPIU8DoJHEC4dyw=
go.opentelemetry.io/otel v1.47.0 h1:j7ALJ/zgkS7Z6aeJW09p8VC9804bC+PpeTfCD4XPnOM=
go.opentelemetry.io/otel v1.47.0/go.mod h1:8wS9O2qfXrYrzp6hIF/HOYJJf/wIhFPhR2xLuP+iXQU=
go.opentelemetry.io/otel/log v1.47.0 h1:cOTS1CcLbSQeZKanGJ+0JpF/+t4PELi3O3bbl2lqCcI=
go.opentelemetry.io/otel/log v1.47.0/go.mod h1:9byitSQ5pLC6PpqwGXjqdMKya6ZTswHRZh2vvXT33nw=
go.opentelemetry.io/otel/metric v1.47.0 h1:4PptaldXx3Eat1XjMZ68pPJEs5wrhlemctZE9a3UdWY=
go.opentelemetry.io/otel/metric v1.47.0/go.mod h1:ADGSXxRrXM6bjbvLo535EstVFlPpPYZm4LBKixjDHwU=
go.opentelemetry.io/otel/sdk v1.47.0 h1:zWXEr4j2lFefG87TU6Yg8a7ngfohIKFZHKp0Hf5hC6I=
go.opentelemetry.io/otel/sdk v1.47.0/go.mod h1:VUc24kiOeoGsxG8G9ULx3fWKvB7jMhnGE8Oi607lgR0=
go.opentelemetry.io/otel/sdk/metric v1.47.0 h1:lfISg2j93VT6yqdk9OfUaZmw/GfcZqCCV3jdXtsPnKw=
go.opentelemetry.io/otel/sdk/metric v1.47.0/go.mod h1:ypLp+mW1Nt2x+Szt3b5/i1syodyts49lMOwxpDI3VGw=
go.opentelemetry.io/otel/trace v1.47.0 h1:JOjX/Oci8K94QHddo+bbfya/Ai/nf6/dt9ZfrFNWSrM=
go.opentelemetry.io/otel/trace v1.47.0/go.mod h1:jNaSLa2PZEYFG6fRjJABAu+bw4FS08uDmPg28lTghu0=
go.uber.org/mock v0.6.0 h1:hyF9dfmbgIX5EfOdasqLsWD6xqpNZlXblLB/Dbnwv3Y=
go.uber.org/mock v0.6.0/go.mod h1:KiVJ4BqZJaMj4svdfmHM0AUx4NJYO8ZNpPnZn1Z+BBU=
go.uber.org/multierr v1.11.0 h1:blXXJkSxSSfBVBlC76pxqeO+LN3aDfLQo+309xJstO0=
go.uber.org/multierr v1.11.0/go.mod h1:20+QtiLqy0Nd6FdQB9TLXag12DsQkrbs3htMFfDN80Y=
go.yaml.in/yaml/v3 v3.0.5 h1:N6y/pJk8buWs9NY5ERU2HSMfm+IuD/OtfdAnq6kESPw=
go.yaml.in/yaml/v3 v3.0.5/go.mod h1:HVTZu1O7/Vkt2N+BFy8Zza+lnLsABggaTM2ZpNIGuKg=
golang.org/x/crypto v0.57.0 h1:3ZVCjf8Ggz7zneR/EHRVx68Ctf+2pmIMP2UFhh9cC6M=
golang.org/x/crypto v0.57.0/go.mod h1:Fdz0i5U6CoizGwLda9DttjSk6qlZo25zYNtR+ycvuZA=
golang.org/x/exp v0.0.0-20260908205506-85c1c2202aba h1:Ck8QetSgk912qxWLMCKxd0in+aiyBQyDSMae6e/xmpU=
golang.org/x/exp v0.0.0-20260908205506-85c1c2202aba/go.mod h1:50RgIsmK7OwqzTTeqcSXQW8SswW0o8fRcDxmqGluJ8E=
golang.org/x/mod v0.41.0 h1:qJmnOUb4YB+FsEuM3HcWucdZASCPGhsX6uljO6pog0c=
golang.org/x/mod v0.41.0/go.mod h1:Ek9pY8RKWXwsWvd3rQiHYtMqkjSUV+s1Rj7j4H5Ur6o=
golang.org/x/net v0.59.0 h1:5zfYln+w5XCxwrnMMJPufRgNoXEaGxl0wo5GqPXyues=
golang.org/x/net v0.59.0/go.mod h1:2DA/G1UfVbCpQPeWTmMPGY7Cs2PkBkwu743bVX5PIVg=
golang.org/x/crypto v0.58.0 h1:COkYLr4k7nDI4r6QtMML2AfzeIswWHAJthLS6K/M/54=
golang.org/x/crypto v0.58.0/go.mod h1:Xh+kl5A+M0gMiWDytH2Hhr5Z7WbNCpv/eB/EFgfSk+w=
golang.org/x/exp v0.0.0-20261013213113-2ab7f3cbc5fb h1:fMK59tNbIKghD6eTnk7DGfVtbK6JiHiFgpmBAWcPaco=
golang.org/x/exp v0.0.0-20261013213113-2ab7f3cbc5fb/go.mod h1:DBoxwWhjaW8DJJw/YeM4zpeOKMC1FMXSblsYy+1mFSw=
golang.org/x/mod v0.42.0 h1:ICeL3t2L30ljSPHHxDnR4fqVcBt/zR2zJAb9juavmQg=
golang.org/x/mod v0.42.0/go.mod h1:u5fzIrRInkL8MaO1cbRkUHT64v303YIT5KEqpNr7jRs=
golang.org/x/net v0.61.0 h1:RR+6j/BTBCrggOLugxa+66R8FvoN6MHlzDlZIbN9YFw=
golang.org/x/net v0.61.0/go.mod h1:WgDCOTH8iwtB66T4/GfI94AHo/hhdb7Hah9NDseQfiI=
golang.org/x/oauth2 v0.37.0 h1:JUlcxA8oAtauLfiH8FX2/FkAWHAdi0QtGCGc+hofE98=
golang.org/x/oauth2 v0.37.0/go.mod h1:IxwZNxUULJmpBFf9K/9NTMSIfZZuvuTy1gGxhigP/58=
golang.org/x/sync v0.23.0 h1:KameEIfc1IkluZyXWLn39Wd4tURc6GbCiISGiZm2bQk=
golang.org/x/sync v0.23.0/go.mod h1:sUUOizhqBxiL6pEWpqNLUiaJn1ShEbZ6BBqskPbjZm0=
golang.org/x/sys v0.0.0-20190916202348-b4ddaad3f8a3/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
golang.org/x/sync v0.24.0 h1:iqiDrGoDgyuvAKrhLQthUEzU/R378N78gXbBjxed7oA=
golang.org/x/sync v0.24.0/go.mod h1:sUUOizhqBxiL6pEWpqNLUiaJn1ShEbZ6BBqskPbjZm0=
golang.org/x/sys v0.0.0-20200116001909-b77594299b42/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
golang.org/x/sys v0.0.0-20200223170610-d5e6a3e2c0ae/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
golang.org/x/sys v0.0.0-20201204225414-ed752295db88/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
golang.org/x/sys v0.0.0-20210630005230-0f9fa26af87c/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
golang.org/x/sys v0.0.0-20210927094055-39ccf1dd6fa6/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
golang.org/x/sys v0.0.0-20220412211240-33da011f77ad/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
golang.org/x/sys v0.0.0-20220503163025-988cb79eb6c6/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
golang.org/x/sys v0.1.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
golang.org/x/sys v0.21.0/go.mod h1:/VUhepiaJMQUp4+oa/7Zr1D23ma6VTLIYjOOTFZPUcA=
golang.org/x/sys v0.48.0 h1:bbX/i/6MgT9BVLM9RT1thmxL04yeTAhbEz4SyadbXoo=
golang.org/x/sys v0.48.0/go.mod h1:hNLxWAXmnKAxqDtdwIYC4bM9oQPEecfsnNMuSxOs3og=
golang.org/x/term v0.46.0 h1:3+OXuTbaKDgwk8jTi3aSLHRlmWqHEUDUtxnbFigO4YE=
golang.org/x/term v0.46.0/go.mod h1:+K02xbkittuwc0Am4abfA3Fc+XRGXkvBXNO88NCXPoc=
golang.org/x/text v0.42.0 h1:JbOZXgfeCPU9gacVtYliJqOhD+zhrEqK4LfdpmlUZqI=
golang.org/x/text v0.42.0/go.mod h1:ojzP1Z+2QtioaF8DTtO8K5q7JWVVYwZKenzujK0Zd0E=
golang.org/x/sys v0.49.0 h1:XbzkgYJHdqh/8m2Uu0W/dQv8nktxx4BFHp1M0gROTXA=
golang.org/x/sys v0.49.0/go.mod h1:hNLxWAXmnKAxqDtdwIYC4bM9oQPEecfsnNMuSxOs3og=
golang.org/x/term v0.47.0 h1:cpyOwQkr2Nzxe2p6KnrTG7Ei2wtzjsuSjhpXXepZ/BM=
golang.org/x/term v0.47.0/go.mod h1:efAfAuG6RKABvg1ArGebRkvX6LXul14SM67o2e4mUCY=
golang.org/x/text v0.43.0 h1:1QivrlhwAsnMOcqgOdl9my7s9OyZl8O5/UGUTJVXoWc=
golang.org/x/text v0.43.0/go.mod h1:hbSIYA/amXcRXTFSZqh8tWHEoqRvvbKPJ2uz/KU+a+s=
golang.org/x/time v0.16.0 h1:vMb6ptszcQMkcwiRTAuNNU50gom6++Q/6gY2hDM6VDE=
golang.org/x/time v0.16.0/go.mod h1:rVKOqvZeKvrDKTQiAHJ7wmwP0RzleSphoEA9RcdLA0s=
golang.org/x/tools v0.50.0 h1:c2ifzfcuY7L90lZ2aKd8S4K2NpASF08SZx9ZuJkHmSU=
golang.org/x/tools v0.50.0/go.mod h1:7ulVMw3831Mwi5EZD6RomGyffr4VFjuNYXf2BbCEAV0=
golang.org/x/tools v0.52.0 h1:ryxg4U63oaUYk7uOLTF37n43f4qgtGvqV4Ioit1BaEs=
golang.org/x/tools v0.52.0/go.mod h1:ItcP5UNaVGG+Yc4eIYmaFLNJ4soEopkx5Y2d4lcEurs=
gonum.org/v1/gonum v0.17.0 h1:VbpOemQlsSMrYmn7T2OUvQ4dqxQXU+ouZFQsZOx50z4=
gonum.org/v1/gonum v0.17.0/go.mod h1:El3tOrEuMpv2UdMrbNlKEh9vd86bmQ6vqIcDwxEOc1E=
google.golang.org/genproto/googleapis/api v0.0.0-20260921155816-b14227669459 h1:GS9OIt/j7c8bvBjYNgnKQysVfmV7e4jM0H8ZK95G4t8=
google.golang.org/genproto/googleapis/api v0.0.0-20260921155816-b14227669459/go.mod h1:PX5/4vemwVoXtwEcRDWwcR1/r0qrosfx3qoVADMwnVE=
google.golang.org/genproto/googleapis/rpc v0.0.0-20260921155816-b14227669459 h1:b0xCahf3FK2m2Cv0p4vTozGPWncCvLfwV86UNg8xWU8=
google.golang.org/genproto/googleapis/rpc v0.0.0-20260921155816-b14227669459/go.mod h1:OaIUM3+LpYcK2GXM4FTmhWoIq371Owdr+Cc7/BsYHHc=
google.golang.org/grpc v1.85.0-dev.0.20260825072537-93e31b48545e h1:ar+feAPij1Znzug45m6AoShUCc5AaR8FLA3I6LZd94k=
google.golang.org/grpc v1.85.0-dev.0.20260825072537-93e31b48545e/go.mod h1:7aY/KlAHanMAVIWHnk8wloQsdlfwzDz/PP9vCbJG6CU=
google.golang.org/genproto/googleapis/api v0.0.0-20261012220534-4341edf3f941 h1:DS/w3N8rsfrwxwgGSj6WhMsPo7aNQGmDx74J81zd0NQ=
google.golang.org/genproto/googleapis/api v0.0.0-20261012220534-4341edf3f941/go.mod h1:AuGAdi9KnXss7xbLxIKhCvk+Ro/SxpET/iS45DNYN+M=
google.golang.org/genproto/googleapis/rpc v0.0.0-20261012220534-4341edf3f941 h1:QG6hS4zFplRZvOQedzei4t2pT/zkQc3VR1u+S9/rcAw=
google.golang.org/genproto/googleapis/rpc v0.0.0-20261012220534-4341edf3f941/go.mod h1:OaIUM3+LpYcK2GXM4FTmhWoIq371Owdr+Cc7/BsYHHc=
google.golang.org/grpc v1.86.0-dev h1:FLq7hcApuJeJ70J9+k0FvmXsNCwS8B2/MeAfYdy3GHY=
google.golang.org/grpc v1.86.0-dev/go.mod h1:Ovl0ECo4xx5r4kn/6d4BPSNB7OIFuu6EAjOzjtVAKaM=
google.golang.org/protobuf v1.36.12 h1:pJOKDDOyeXErUroCihFAd5LQuwXBSpVnKGrj5o/fwxc=
google.golang.org/protobuf v1.36.12/go.mod h1:HTf+CrKn2C3g5S8VImy6tdcUvCska2kB7j23XfzDpco=
gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=
@@ -319,15 +277,15 @@ gopkg.in/yaml.v3 v3.0.1 h1:fxVm/GzAzEWqLHuvctI91KS9hhNmmWOoWu0XTYJS7CA=
gopkg.in/yaml.v3 v3.0.1/go.mod h1:K4uyk7z7BCEPqu6E+C64Yfv1cQ7kz7rIZviUmN+EgEM=
gotest.tools/v3 v3.5.2 h1:7koQfIKdy+I8UTetycgUqXWSDwpgv193Ka+qRsmBY8Q=
gotest.tools/v3 v3.5.2/go.mod h1:LtdLGcnqToBH83WByAAi/wiwSFCArdFIUV/xxN4pcjA=
modernc.org/libc v1.77.1 h1:Ct8j47QtiZ1Enj2DtFXQtUqrPCAjdCmPjtCuvrYQ0Hs=
modernc.org/libc v1.77.1/go.mod h1:87/pZ4L6nD1zqW4nItuS12YO7hN1igAah34xjnQo/W0=
modernc.org/libc v1.78.0 h1:Lpm+e9K6ruVYEGwWfuBO1xXtYEaXDouFOG1SuWjFDP4=
modernc.org/libc v1.78.0/go.mod h1:wvxR55nOF2/66NkJoNV7/coXBTCvT3HR22rcqW6lOuA=
modernc.org/mathutil v1.7.1 h1:GCZVGXdaN8gTqB1Mf/usp1Y/hSqgI2vAGGP4jZMCxOU=
modernc.org/mathutil v1.7.1/go.mod h1:4p5IwJITfppl0G4sUEDtCr4DthTaT47/N3aT6MhfgJg=
modernc.org/memory v1.12.1 h1:nFMiWrpStgZczNl6XI9GnIk/rWhYIyHGUaR04pGbp9g=
modernc.org/memory v1.12.1/go.mod h1:/JP4VbVC+K5sU2wZi9bHoq2MAkCnrt2r98UGeSK7Mjw=
modernc.org/sqlite v1.59.0 h1:X1es1GpqBlS/5T+vbM4HLUdaa8OtQx468DF2vrx+38A=
modernc.org/sqlite v1.59.0/go.mod h1:+paeT2A3iPRHkQDwG7oA6Tk0zQd5woMEI8q7orfry8k=
oras.land/oras-go/v2 v2.6.2 h1:N04RXngAp1LJKTG6ifz3xHPipasEkWr+hFmInja5YKo=
oras.land/oras-go/v2 v2.6.2/go.mod h1:PlTtg4JTDJkDe8yVHpM2wz7/YDc00GVas+i4jAW2TZ4=
modernc.org/sqlite v1.60.1 h1:/blz53O951KWFOso4QQvEs/Fq6cDBKLtMVrYNSeJVKw=
modernc.org/sqlite v1.60.1/go.mod h1:1dIoEagfDE72QytD5scH1lxARtaUgKgHC/NuApA27r0=
oras.land/oras-go/v2 v2.6.3 h1:72z4K9YiDKVB5jieM8cuB1WSVnbZkkD2hd+NmO3HYVI=
oras.land/oras-go/v2 v2.6.3/go.mod h1:imXEFiYkTZUVVYkfv/r5GVmWeiLkGtNQA8FU4F/sjKs=
pgregory.net/rapid v1.2.0 h1:keKAYRcjm+e1F0oAuU5F5+YPAWcyxNNRK2wud503Gnk=
pgregory.net/rapid v1.2.0/go.mod h1:PY5XlDGj0+V1FCq0o192FdRhpKHGTRIWBgqjDBTrq04=
+4 -5
View File
@@ -89,8 +89,8 @@ fi
mkdir -p "${ARCHIE_CONFIG_DIR}" "${ARCHIE_DATA_DIR}" "${ARCHIE_BIN_DIR}"
# 2. Binaries. STAGE ends up holding the binaries, release.json and the
# updater scripts, whichever way they were obtained.
# 2. Binaries. STAGE ends up holding the binaries and release.json, whichever
# way they were obtained.
TMP="$(mktemp -d)"
trap 'rm -rf "${TMP}"' EXIT
@@ -111,7 +111,7 @@ if [ "${FROM_SOURCE}" = true ]; then
for cmd in $(jq -er '.required_topology.units | join(" ")' "${SRC_DIR}/release.json"); do
(cd "${SRC_DIR}" && go build -ldflags "${LDFLAGS}" -o "${STAGE}/${cmd}" "./cmd/${cmd}")
done
cp "${SRC_DIR}/release.json" "${SRC_DIR}"/scripts/archie-update-* "${STAGE}/"
cp "${SRC_DIR}/release.json" "${STAGE}/"
else
if [ -z "${VERSION}" ]; then
tag="$(curl -fsSL "https://api.github.com/repos/${REPO}/releases/latest" | sed -n 's/.*"tag_name": *"\([^"]*\)".*/\1/p')"
@@ -137,8 +137,7 @@ TOPOLOGY_UNITS="$(tr -d '\n' <"${STAGE}/release.json" | sed -n 's/.*"units": *\[
for cmd in ${TOPOLOGY_UNITS}; do
install -m755 "${STAGE}/${cmd}" "${ARCHIE_BIN_DIR}/${cmd}"
done
install -m755 "${STAGE}"/archie-update-* "${ARCHIE_BIN_DIR}/"
echo " Installed ${TOPOLOGY_UNITS} and the updater to ${ARCHIE_BIN_DIR}/"
echo " Installed ${TOPOLOGY_UNITS} to ${ARCHIE_BIN_DIR}/"
# 3. Configuration. archied setup asks every question and writes config.toml
# plus any secrets to the env file beside it; this script asks nothing itself.
+5 -3
View File
@@ -254,16 +254,18 @@ func invoke(ctx context.Context, workspace string, h *agentrun.HarnessSpec, argv
return err
}
var stderr tailBuffer
cmd.Stderr = &stderr
cmd.Stderr = io.MultiWriter(&stderr, os.Stderr)
if err := cmd.Start(); err != nil {
return fmt.Errorf("start harness: %w", err)
}
scanner := bufio.NewScanner(stdout)
// Keep the raw harness activity visible in the container's live output.
output := io.TeeReader(stdout, os.Stdout)
scanner := bufio.NewScanner(output)
scanner.Buffer(make([]byte, 64*1024), 16*1024*1024)
for scanner.Scan() {
out.Line(scanner.Bytes(), report)
}
_, _ = io.Copy(io.Discard, stdout)
_, _ = io.Copy(io.Discard, output)
if err := cmd.Wait(); err != nil {
if ctx.Err() != nil {
return ctx.Err()
+41
View File
@@ -0,0 +1,41 @@
//go:build unix
package agentexec
import (
"os"
"os/exec"
"strings"
"testing"
"github.com/samcharles93/archie-core/internal/domain/agentrun"
)
func TestHarnessLiveOutput(t *testing.T) {
if os.Getenv("ARCHIE_TEST_HARNESS_OUTPUT") == "1" {
adapter := &codexHarnessOutput{}
user := ""
if os.Geteuid() == 0 {
user = "65534"
}
err := invoke(t.Context(), "/tmp", &agentrun.HarnessSpec{User: user}, []string{"sh", "-c", `printf '%s\n' '{"type":"thread.started","thread_id":"live-thread"}'; printf '%s\n' 'live diagnostic' >&2`}, adapter, nil)
if err != nil {
t.Fatal(err)
}
if adapter.SessionID() != "live-thread" {
t.Fatal("forwarding lost the adapter input")
}
return
}
cmd := exec.CommandContext(t.Context(), os.Args[0], "-test.run=^TestHarnessLiveOutput$")
cmd.Env = append(os.Environ(), "ARCHIE_TEST_HARNESS_OUTPUT=1")
got, err := cmd.CombinedOutput()
if err != nil {
t.Fatalf("harness: %v: %s", err, got)
}
for _, want := range []string{"live-thread", "live diagnostic"} {
if !strings.Contains(string(got), want) {
t.Errorf("worker output %q is missing %q", got, want)
}
}
}
+46 -1
View File
@@ -8,6 +8,8 @@ import (
"slices"
"strings"
"github.com/google/jsonschema-go/jsonschema"
"github.com/samcharles93/archie-core/internal/domain/agentrun"
)
@@ -62,7 +64,7 @@ func ProtectionMatcher(p agentrun.Protection, readOnly bool) func(string) bool {
func ValidateCaptureArgs(spec agentrun.CaptureTool, value json.RawMessage) (string, bool) {
var object map[string]json.RawMessage
if err := json.Unmarshal(value, &object); err != nil {
return spec.Name + " rejected: arguments must be a JSON object", false //nolint:nilerr // the agent loop must see malformed tool arguments as feedback it can correct, not as a failed tool call
return spec.Name + " rejected: arguments must be a JSON object", false // the agent loop must see malformed tool arguments as feedback it can correct, not as a failed tool call
}
for _, check := range []func(agentrun.CaptureTool, map[string]json.RawMessage) (string, bool){
checkRequiredFields,
@@ -74,9 +76,52 @@ func ValidateCaptureArgs(spec agentrun.CaptureTool, value json.RawMessage) (stri
return rejection, false
}
}
// The derived checks above cover the common cases with friendly wording;
// the schema is what enforces the rest of what the tool advertised --
// enums, numeric types, nested objects, array shapes -- so a call the
// agent could only make valid is rejected with the reason, not recorded.
return checkSchema(spec, value)
}
// checkSchema validates the arguments against the tool's advertised JSON
// Schema. A tool with no schema, or one that will not resolve, is accepted
// here: a schema that never resolves is a definition bug, refused where the
// tool is built (workflow's agentResultTool), so the agent is never blamed
// for it.
func checkSchema(spec agentrun.CaptureTool, value json.RawMessage) (string, bool) {
if len(spec.Parameters) == 0 {
return "", true
}
resolved, err := resolveCaptureSchema(spec.Parameters)
if err != nil {
return "", true
}
var instance any
if err := json.Unmarshal(value, &instance); err != nil {
return spec.Name + " rejected: arguments must be JSON", false
}
if err := resolved.Validate(instance); err != nil {
return fmt.Sprintf("%s rejected: %v", spec.Name, err), false
}
return "", true
}
// resolveCaptureSchema parses and resolves a capture tool's JSON Schema. The
// workflow parser resolves the same schema when it builds the tool, so an
// unusable schema is refused at definition time rather than silently skipped
// here.
func resolveCaptureSchema(params json.RawMessage) (*jsonschema.Resolved, error) {
var schema jsonschema.Schema
if err := json.Unmarshal(params, &schema); err != nil {
return nil, fmt.Errorf("capture tool schema: %w", err)
}
resolved, err := schema.Resolve(nil)
if err != nil {
return nil, fmt.Errorf("capture tool schema: %w", err)
}
return resolved, nil
}
func checkRequiredFields(spec agentrun.CaptureTool, object map[string]json.RawMessage) (string, bool) {
for _, field := range spec.RequiredFields {
if _, ok := object[field]; !ok {
+98
View File
@@ -0,0 +1,98 @@
package agentexec
import (
"encoding/json"
"strings"
"testing"
"github.com/samcharles93/archie-core/internal/domain/agentrun"
)
// resultSchema is the shape of a shipped planner result (triage's), with a
// nested object and an array so every constraint kind the agent can violate
// is exercised.
const resultSchema = `{
"type": "object",
"properties": {
"needs_code_change": {"type": "boolean"},
"workflow": {"type": "string", "enum": ["implement", "tdd", "feasibility"]},
"reasons": {"type": "string"},
"labels": {"type": "array", "items": {"type": "string"}, "maxItems": 2},
"plan": {"type": "object", "properties": {"risk": {"type": "integer"}}, "required": ["risk"]}
},
"required": ["needs_code_change", "workflow", "reasons"]
}`
// TestValidateCaptureArgsEnforcesSchema pins that the advertised schema is
// enforced, not merely offered to the agent: an enum, a numeric type, an
// array's item type and length, and a nested object's shape all reject with a
// message naming what was wrong, so the agent can correct it.
func TestValidateCaptureArgsEnforcesSchema(t *testing.T) {
base := `"needs_code_change": true, "workflow": "tdd", "reasons": "why"`
tests := []struct {
name string
args string
wantOK bool
wantText string
}{
{name: "valid", args: `{` + base + `}`, wantOK: true},
{name: "enum violation", args: `{` + base + `, "workflow": "banana"}`, wantText: "workflow"},
{name: "wrong boolean type", args: `{"needs_code_change": "yes", "workflow": "tdd", "reasons": "why"}`, wantText: "needs_code_change"},
{name: "string where integer expected", args: `{` + base + `, "plan": {"risk": "high"}}`, wantText: "plan"},
{name: "array item type", args: `{` + base + `, "labels": [1, 2]}`, wantText: "labels"},
{name: "array too long", args: `{` + base + `, "labels": ["a", "b", "c"]}`, wantText: "labels"},
{name: "nested required missing", args: `{` + base + `, "plan": {}}`, wantText: "plan"},
}
spec := agentrun.CaptureTool{Name: "result", Parameters: json.RawMessage(resultSchema)}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
rejection, ok := ValidateCaptureArgs(spec, json.RawMessage(tt.args))
if ok != tt.wantOK {
t.Fatalf("ok = %v, rejection = %q; want ok %v", ok, rejection, tt.wantOK)
}
if ok {
return
}
if !strings.HasPrefix(rejection, "result rejected: ") {
t.Fatalf("rejection %q does not name the tool", rejection)
}
if !strings.Contains(rejection, tt.wantText) {
t.Fatalf("rejection %q does not mention %q", rejection, tt.wantText)
}
})
}
}
// TestValidateCaptureArgsWithoutSchema pins that a tool with no schema, or one
// whose schema cannot resolve, still accepts a call: the derived constraint
// lists remain the whole check, so the agent is never blocked by a tool
// definition the schema check cannot read.
func TestValidateCaptureArgsWithoutSchema(t *testing.T) {
tests := []struct {
name string
spec agentrun.CaptureTool
args string
wantOK bool
}{
{name: "no schema", spec: agentrun.CaptureTool{Name: "note"}, args: `{"anything": 1}`, wantOK: true},
{
name: "unresolvable schema is not the agent's problem",
spec: agentrun.CaptureTool{Name: "note", Parameters: json.RawMessage(`{"$ref": "#/definitions/missing"}`)},
args: `{"anything": 1}`,
wantOK: true,
},
{
name: "derived checks still apply without a schema",
spec: agentrun.CaptureTool{Name: "note", NonEmptyStrings: []string{"body"}},
args: `{"body": " "}`,
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
_, ok := ValidateCaptureArgs(tt.spec, json.RawMessage(tt.args))
if ok != tt.wantOK {
t.Fatalf("ok = %v, want %v", ok, tt.wantOK)
}
})
}
}
+3 -3
View File
@@ -31,7 +31,7 @@ type LoopRunner struct {
// Limits bounds the size of tool results fed back into the stage. The
// zero value applies no limits.
Limits ToolLimits
// AllowTools is the profile's allowlist over archie-added tools. Empty allows
// AllowTools is the allowlist over archie-added tools. Nil allows
// all. Capture tools and agentloop's built-in file tools are never
// filtered.
AllowTools []string
@@ -187,9 +187,9 @@ func makeCaptureHandler(spec agentrun.CaptureTool, captures map[string][]json.Ra
}
}
// allowedTools keeps the tools allow names; an empty allow keeps them all.
// allowedTools keeps the tools allow names; a nil allow keeps them all.
func allowedTools(set core.ToolSet, allow []string) core.ToolSet {
if len(allow) == 0 {
if allow == nil {
return set
}
kept := core.ToolSet{}
+30 -11
View File
@@ -4,31 +4,31 @@ import (
"context"
"errors"
"fmt"
"strings"
"log/slog"
"time"
"github.com/samcharles93/ai-sdk/chat"
"github.com/samcharles93/ai-sdk/core"
"github.com/samcharles93/ai-sdk/runtime"
"github.com/samcharles93/archie-core/internal/agentexec/modelloop"
"github.com/samcharles93/archie-core/internal/config"
"github.com/samcharles93/archie-core/internal/domain/usage"
"github.com/samcharles93/archie-core/internal/domain/workflow"
"github.com/samcharles93/archie-core/internal/tools/mcp"
)
// taskSamplingModel returns the task's builder model, or "".
func taskSamplingModel(cfg config.TaskConfig) string {
return strings.TrimSpace(cfg.Models["builder"])
}
// taskSamplingHandler answers MCP sampling requests with the task's model.
// Without a runtime or model the request is refused.
func taskSamplingHandler(llm *runtime.Runtime, cfg config.TaskConfig) mcp.SamplingHandler {
// Without a runtime or model the request is refused. record receives each
// call's model ref and token usage.
func taskSamplingHandler(llm *runtime.Runtime, cfg config.TaskConfig, record func(context.Context, string, chat.Usage)) mcp.SamplingHandler {
return func(ctx context.Context, req mcp.SamplingRequest) (mcp.SamplingResult, error) {
if llm == nil {
return mcp.SamplingResult{}, errors.New("mcp sampling: no model runtime is configured")
}
model := taskSamplingModel(cfg)
if model == "" {
return mcp.SamplingResult{}, errors.New("mcp sampling: no builder model is configured")
model, err := config.ResolveModel(cfg.Models, "", config.PurposeAgent)
if err != nil {
return mcp.SamplingResult{}, fmt.Errorf("mcp sampling: %w", err)
}
messages, err := modelloop.SamplingMessages(req)
if err != nil {
@@ -45,6 +45,7 @@ func taskSamplingHandler(llm *runtime.Runtime, cfg config.TaskConfig) mcp.Sampli
options.Temperature = float32(*req.Temperature)
}
result, err := llm.Chat(ctx, model, options)
record(ctx, model, result.TotalUsage)
if err != nil {
return mcp.SamplingResult{}, fmt.Errorf("mcp sampling: %w", err)
}
@@ -56,3 +57,21 @@ func taskSamplingHandler(llm *runtime.Runtime, cfg config.TaskConfig) mcp.Sampli
}, nil
}
}
// samplingUsageRecorder records a task's MCP sampling calls against the task.
// A failed write is logged, never fails the call.
func samplingUsageRecorder(store workflow.Store, t *workflow.Task, log *slog.Logger) func(context.Context, string, chat.Usage) {
return func(ctx context.Context, model string, used chat.Usage) {
if store == nil || t == nil {
return
}
record := usage.Record{
Source: usage.SourceTask, TaskID: t.ID, Attempt: t.Attempt, Workflow: t.Workflow, Step: "mcp-sampling",
InputTokens: int64(used.PromptTokens), OutputTokens: int64(used.CompletionTokens),
CachedTokens: int64(used.CachedTokens), At: time.Now().UTC(),
}.WithRef(model)
if err := store.RecordUsage(ctx, record); err != nil && log != nil {
log.Warn("record model usage failed", "model", model, "err", err)
}
}
}
+14 -7
View File
@@ -198,8 +198,6 @@ func stageRunners(req taskrun.Request, mcpSet *mcpProviderSet, newRunner runnerF
// routeTask remains available for routing-only callers. Production execution
// compiles the request's pinned YAML directly.
func routeTask(req taskrun.Request, registry workflow.Registry) workflow.Workflow {
workflow.SetKindWorkflows(req.KindWorkflows)
workflow.SetLabelWorkflows(req.LabelWorkflows)
return workflow.Route(req.Task, registry)
}
@@ -233,7 +231,7 @@ func runTask(ctx context.Context, req taskrun.Request, dependencies taskDependen
if len(req.MCPServers) > 0 {
llm = modelloop.NewRuntime(req.Providers, req.Cfg.ModelLimits)
}
mcpSet, mcpErr := startMCPProviders(ctx, req.MCPServers, taskSamplingHandler(llm, req.Cfg), log)
mcpSet, mcpErr := startMCPProviders(ctx, req.MCPServers, taskSamplingHandler(llm, req.Cfg, samplingUsageRecorder(dependencies.store, req.Task, log)), log)
if mcpSet != nil {
defer mcpSet.cleanup(ctx, log)
}
@@ -269,10 +267,7 @@ func runTask(ctx context.Context, req taskrun.Request, dependencies taskDependen
Log: log,
Guardrails: dependencies.guardrails,
}
// pr-review reads the PR data the daemon prefetched into workDir.
if req.Task.Workflow == "pr-review" {
tc.PRSource = prsource.NewFromMount(filepath.Join(workDir, prsource.PrefetchDirName))
}
configureTaskContext(tc, workDir)
if req.Repo.PersistentStorage {
memory, readErr := readProjectMemory(storage.MemoryPath)
@@ -294,6 +289,18 @@ func runTask(ctx context.Context, req taskrun.Request, dependencies taskDependen
}, nil
}
// configureTaskContext applies what the task's kind changes about its context:
// a test run writes nothing to a forge, and pr-review reads the PR data the
// daemon prefetched into workDir.
func configureTaskContext(tc *workflow.TaskContext, workDir string) {
if tc.Task.IsTestRun() {
workflow.SuppressForgeWrites(tc)
}
if tc.Task.Workflow == "pr-review" {
tc.PRSource = prsource.NewFromMount(filepath.Join(workDir, prsource.PrefetchDirName))
}
}
// CompilePinnedWorkflow compiles the definition the request carries. With
// none, it pins one from the shipped definitions onto req and req.Task.
func CompilePinnedWorkflow(req *taskrun.Request, steps *workflow.Manager) (workflow.Workflow, error) {
+38 -5
View File
@@ -3,16 +3,25 @@ package agentworker
import (
"context"
"errors"
"log/slog"
"net/http"
"strconv"
"time"
"github.com/samcharles93/archie-core/internal/agentexec"
"github.com/samcharles93/archie-core/internal/domain/health"
"github.com/samcharles93/archie-core/internal/domain/presence"
"github.com/samcharles93/archie-core/internal/domain/storecontract"
"github.com/samcharles93/archie-core/internal/domain/workflow"
"github.com/samcharles93/archie-core/internal/domain/workflow/task"
"github.com/samcharles93/archie-core/internal/infrastructure/agentboot"
"github.com/samcharles93/archie-core/internal/infrastructure/agentgit"
agentnats "github.com/samcharles93/archie-core/internal/infrastructure/agenttransport/nats"
"github.com/samcharles93/archie-core/internal/infrastructure/egress"
"github.com/samcharles93/archie-core/internal/infrastructure/readiness"
"github.com/samcharles93/archie-core/internal/infrastructure/workflowsteps"
"github.com/samcharles93/archie-core/internal/installtype"
"github.com/samcharles93/archie-core/internal/storage"
"github.com/samcharles93/archie-core/internal/taskrun"
"github.com/samcharles93/archie-core/internal/tools"
@@ -41,17 +50,24 @@ type Settings struct {
// NATS storerpc path is deleted.
StateStoreTarget string
StateStoreToken string
// StateStoreCA is the PEM CA the State Store's certificate chains to.
StateStoreCA []byte
// ModelProxy, ModelProxyHosts and ModelProxyCA route model provider
// requests through the daemon's egress proxy, which holds the keys.
// Empty ModelProxy sends them directly.
ModelProxy string
ModelProxyHosts []string
ModelProxyCA string
}
type workerTransport interface {
taskServiceTransport
Close()
LogPublisher() agentexec.LogPublisher
EventPublisher() agentexec.EventPublisher
SubscribeTasks(context.Context, int64, agentnats.TaskHandler, *slog.Logger) (agentnats.Subscription, error)
Forger(identity, credential string, timeout time.Duration) workflow.Forger
Store(time.Duration) workflow.Store
Calls(time.Duration) task.Caller
Trees(string, string, time.Duration) agentnats.RemoteTrees
PutPresence(context.Context, storecontract.Presence) error
BrokerReady() bool
}
type workerDependencies struct {
@@ -110,6 +126,15 @@ func run(ctx context.Context, settings Settings, log *slog.Logger, dependencies
workDir = storage.WorktreeMountDir
}
if settings.ModelProxy != "" {
transport, err := egress.ModelTransport(settings.ModelProxy, settings.ModelProxyHosts, settings.ModelProxyCA)
if err != nil {
return &StartupError{Operation: "model proxy", Err: err}
}
// The model SDK builds its clients on the default transport and takes
// no client of its own, so the process-wide default is the seam.
http.DefaultTransport = transport
}
dependencies.markSafe(ctx, workDir, log)
// The task is known before connecting: without a broker token the
@@ -125,6 +150,7 @@ func run(ctx context.Context, settings Settings, log *slog.Logger, dependencies
TaskID: taskID,
StateStoreURL: settings.StateStoreTarget,
StateStoreToken: settings.StateStoreToken,
StateStoreCA: settings.StateStoreCA,
}, log)
if err != nil {
operation := "nats connect failed"
@@ -148,6 +174,13 @@ func run(ctx context.Context, settings Settings, log *slog.Logger, dependencies
}
}()
registry := health.NewRegistry(readiness.NewContractProbe("nats", 0, func(context.Context) error {
if !transport.BrokerReady() {
return errors.New("broker disconnected")
}
return nil
}))
go presence.Run(ctx, presence.Agent, presence.Build{InstanceID: strconv.FormatInt(taskID, 10), Version: Version(), InstallType: installtype.Type()}, transport, registry, log)
log.Info("archie-agent ready", "nats", settings.NATSURL)
dependencies.wait(ctx, log)
return nil
+349
View File
@@ -0,0 +1,349 @@
package archied
import (
"context"
"fmt"
"maps"
"net/http"
"slices"
"strings"
"sync"
"time"
"github.com/samcharles93/archie-core/internal/config"
"github.com/samcharles93/archie-core/internal/domain/alert"
"github.com/samcharles93/archie-core/internal/domain/org"
"github.com/samcharles93/archie-core/internal/infrastructure/alertwebhook"
"github.com/samcharles93/archie-core/internal/infrastructure/controlplanerpc"
"github.com/samcharles93/archie-core/internal/infrastructure/modelcatalog"
"github.com/samcharles93/archie-core/internal/releaseupdate"
)
const (
// deliveryInterval is how often due webhook deliveries are claimed.
deliveryInterval = 15 * time.Second
// releaseCheckInterval is how often GitHub is asked for a newer release.
releaseCheckInterval = 6 * time.Hour
// alertCallTimeout bounds one check's calls to the State Store.
alertCallTimeout = 20 * time.Second
// credentialReportEvery is how long an unchanged credential state goes
// unreported, so the egress path does not call the State Store per request.
credentialReportEvery = 10 * time.Minute
)
// alertChecks raises the alerts that have no transition to hang on: the
// daemon's configuration and the release channel. Each check lists what it
// owns, so alerts from before a restart are still cleared once fixed.
type alertChecks struct {
store alert.Store
log func(msg string, args ...any)
cfg func() config.Config
// catalog is the models the providers currently offer.
catalog func() modelcatalog.Snapshot
// release is the newest release the update channel offers.
release func(context.Context) (releaseupdate.Snapshot, error)
}
func (c alertChecks) reporter() alert.Reporter {
return alert.Reporter{Raiser: c.store, By: "archied"}
}
// sync makes the open alerts of kind in orgID match active: it raises each
// input and resolves the alerts, snoozed or open, whose condition cleared.
func (c alertChecks) sync(ctx context.Context, orgID org.OrgID, kind string, active []alert.Input) {
reporter := c.reporter()
subjects := make([]string, 0, len(active))
for _, in := range active {
reporter.Raise(ctx, in)
subjects = append(subjects, in.Subject)
}
existing, err := c.store.ListAlerts(ctx, orgID, alert.Filter{States: []alert.State{alert.StateOpen, alert.StateDismissed}, Limit: 500})
if err != nil {
c.log("alerts not checked", "kind", kind, "err", err)
return
}
for _, a := range existing {
if a.Kind == kind && a.Condition && !slices.Contains(subjects, a.Subject) {
reporter.Resolve(ctx, a.Key)
}
}
}
// checkConfig raises the alerts the effective configuration implies: a
// missing or dangling model alias, and providers and granted services with no
// usable credential.
func (c alertChecks) checkConfig(ctx context.Context) {
cfg := c.cfg()
c.sync(ctx, org.DefaultOrgID, alert.KindModelAlias, modelAliasAlerts(cfg.Models, c.catalog()))
c.sync(ctx, org.DefaultOrgID, alert.KindProviderNoKey, providerAlerts(cfg.Providers))
unbound := unboundAlerts(cfg)
for _, orgID := range credentialOrgs(cfg) {
c.sync(ctx, orgID, alert.KindToolNoBinding, unbound[orgID])
}
}
func modelAliasAlerts(aliases map[string]string, catalog modelcatalog.Snapshot) []alert.Input {
var out []alert.Input
input := func(alias, title, body string) alert.Input {
return alert.Input{
Org: org.DefaultOrgID, Kind: alert.KindModelAlias, Subject: "alias:" + alias,
Severity: alert.SeverityAction, Condition: true, Title: title, Body: body, Link: "/settings/models",
}
}
if strings.TrimSpace(aliases[config.DefaultModelAlias]) == "" {
out = append(out, input(config.DefaultModelAlias, "No default model alias is set",
"Set it to a model."))
}
for _, alias := range slices.Sorted(maps.Keys(aliases)) {
providerID, modelID, ok := strings.Cut(aliases[alias], "/")
if !ok {
continue
}
offered := catalogModels(catalog, providerID)
// A provider the catalogue does not list is a credential problem, not
// a retired model.
if len(offered) > 0 && !slices.Contains(offered, modelID) {
out = append(out, input(alias, fmt.Sprintf("Model alias %q points at a model that no longer exists", alias),
fmt.Sprintf("The %s provider no longer offers the model this alias names. Choose another model.", providerID)))
}
}
return out
}
func catalogModels(catalog modelcatalog.Snapshot, providerID string) []string {
for _, provider := range catalog.Providers {
if provider.ID == providerID {
ids := make([]string, len(provider.Models))
for i, model := range provider.Models {
ids[i] = model.ID
}
return ids
}
}
return nil
}
func providerAlerts(providers map[string]config.Provider) []alert.Input {
var out []alert.Input
for _, name := range slices.Sorted(maps.Keys(providers)) {
if !providers[name].Disabled {
continue
}
out = append(out, alert.Input{
Org: org.DefaultOrgID, Kind: alert.KindProviderNoKey, Subject: "provider:" + name,
Severity: alert.SeverityAction, Condition: true,
Title: fmt.Sprintf("Provider %q has no usable credential", name),
Body: "Key unresolved; its models cannot run. Check the binding.",
Link: "/settings/credential-bindings",
})
}
return out
}
// credentialOrgs are the orgs whose identities are granted credentials, the
// instance org always among them.
func credentialOrgs(cfg config.Config) []org.OrgID {
orgs := []org.OrgID{org.DefaultOrgID}
for _, name := range append([]string{""}, identityNames(cfg)...) {
orgName, _ := cfg.CredentialAccess(name)
if id := org.OrgID(credentialOrg(orgName)); !slices.Contains(orgs, id) {
orgs = append(orgs, id)
}
}
return orgs
}
func identityNames(cfg config.Config) []string {
names := make([]string, 0, len(cfg.Identities))
for _, id := range cfg.Identities {
names = append(names, id.Name)
}
return names
}
func credentialOrg(name string) string {
if name == "" {
return string(org.DefaultOrgID)
}
return name
}
// unboundAlerts are the granted services with no credential binding in the
// granting identity's org, by org. A model provider needs no binding of its
// own here: its key is the provider's.
func unboundAlerts(cfg config.Config) map[org.OrgID][]alert.Input {
out := map[org.OrgID][]alert.Input{}
seen := map[alert.Key]bool{}
for _, name := range append([]string{""}, identityNames(cfg)...) {
orgName, granted := cfg.CredentialAccess(name)
orgID := org.OrgID(credentialOrg(orgName))
bound := cfg.Containers.BoundCredentials(orgName, granted, granted)
for _, service := range granted {
key := alert.Key{Org: orgID, Kind: alert.KindToolNoBinding, Subject: "service:" + service}
if _, ok := cfg.Providers[service]; ok || bound[service].Service != "" || seen[key] {
continue
}
seen[key] = true
out[orgID] = append(out[orgID], alert.Input{
Key: key, Severity: alert.SeverityAction, Condition: true,
Title: fmt.Sprintf("%q is granted but has no credential binding", service),
Body: "Runs cannot authenticate. Bind a credential.",
Link: "/settings/credential-bindings",
})
}
}
return out
}
// checkRelease raises an info alert while the update channel offers a newer
// release. A failed check changes nothing: the next one decides.
func (c alertChecks) checkRelease(ctx context.Context) {
snapshot, err := c.release(ctx)
if err != nil {
c.log("release check failed", "err", err)
return
}
var active []alert.Input
if available := snapshot.Available(); len(available) > 0 {
parts := make([]string, len(available))
for i, component := range available {
parts[i] = fmt.Sprintf("%s %s", component.Label, component.Available)
}
active = append(active, alert.Input{
Org: org.DefaultOrgID, Kind: alert.KindReleaseAvailable, Subject: "release",
Severity: alert.SeverityInfo, Condition: true,
Title: "Newer release available", Body: "Available: " + strings.Join(parts, ", ") + ".",
Link: "/settings/status",
})
}
c.sync(ctx, org.DefaultOrgID, alert.KindReleaseAvailable, active)
}
// every runs fn now and then each interval until ctx ends.
func every(ctx context.Context, interval time.Duration, fn func(context.Context)) {
ticker := time.NewTicker(interval)
defer ticker.Stop()
for {
callCtx, cancel := context.WithTimeout(ctx, alertCallTimeout)
fn(callCtx)
cancel()
select {
case <-ctx.Done():
return
case <-ticker.C:
}
}
}
// startAlerts starts the daemon's alert checks and the webhook delivery loop.
// The State Store being away costs the checks a tick, nothing else.
func (b *boot) startAlerts(ctx context.Context) {
client, ok := b.stateStore.(alert.Store)
if !ok {
return
}
reader := controlplanerpc.NewRPCClient(b.controlPlaneRPC)
catalog := controlplanerpc.UpdateService(reader, "").Catalog
checks := alertChecks{
store: client, log: b.log.Warn, cfg: b.cfgHolder.Get, catalog: b.catalogState,
release: func(ctx context.Context) (releaseupdate.Snapshot, error) { return catalog.Check(ctx) },
}
b.alertChecks = &checks
go every(ctx, releaseCheckInterval, checks.checkRelease)
go every(ctx, deliveryInterval, b.deliverAlerts(client))
checks.checkConfig(ctx)
}
// recheckAlerts re-evaluates the configuration alerts after the configuration
// changed. It never blocks the change that caused it.
func (b *boot) recheckAlerts(ctx context.Context) {
checks := b.alertChecks
if checks == nil {
return
}
go func() {
ctx, cancel := context.WithTimeout(context.WithoutCancel(ctx), alertCallTimeout)
defer cancel()
checks.checkConfig(ctx)
}()
}
// deliverAlerts sends due webhook deliveries. A delivery the webhook is not
// configured for, or whose alert is below the severity threshold, is closed
// as skipped; one that fails is retried by the store on its schedule.
func (b *boot) deliverAlerts(store alert.Deliverer) func(context.Context) {
client := &http.Client{Timeout: 15 * time.Second}
return func(ctx context.Context) {
due, err := store.ClaimAlertDeliveries(ctx, 20)
if err != nil {
b.log.Warn("alert deliveries not claimed", "err", err)
return
}
notify := b.cfgHolder.Get().Notify
for _, d := range due {
outcome := alert.Outcome{DeliveryID: d.ID}
threshold := alert.Severity(notify.MinSeverity)
if !threshold.Valid() {
threshold = alert.SeverityWarning
}
if notify.Webhook == "" || d.Alert.Severity.Rank() < threshold.Rank() {
outcome.Skipped = true
} else {
secret := ""
if notify.Secret != (config.SecretRef{}) {
secret, err = b.secrets.Resolve(notify.Secret)
}
if err != nil {
outcome.Error = "notify secret could not be resolved"
} else {
outcome = alertwebhook.Send(ctx, client, notify.Webhook, secret, d)
}
}
if err := store.RecordAlertAttempt(ctx, outcome); err != nil {
b.log.Warn("alert delivery not recorded", "delivery", d.ID, "err", err)
}
}
}
}
// credentialWatch raises an alert when a bound credential cannot be resolved
// and clears it once it can. The egress path calls observe per request, so an
// unchanged state is reported at most every credentialReportEvery.
type credentialWatch struct {
reporter alert.Reporter
mu sync.Mutex
last map[alert.Key]credentialState
}
type credentialState struct {
rejected bool
at time.Time
}
func (w *credentialWatch) observe(ctx context.Context, orgName, service string, rejected bool) {
if w == nil {
return
}
key := alert.Key{Org: org.OrgID(credentialOrg(orgName)), Kind: alert.KindCredentialInvalid, Subject: "service:" + service}
w.mu.Lock()
if w.last == nil {
w.last = map[alert.Key]credentialState{}
}
previous, known := w.last[key]
if known && previous.rejected == rejected && time.Since(previous.at) < credentialReportEvery {
w.mu.Unlock()
return
}
w.last[key] = credentialState{rejected: rejected, at: time.Now()}
w.mu.Unlock()
go func() {
ctx, cancel := context.WithTimeout(context.WithoutCancel(ctx), alertCallTimeout)
defer cancel()
w.reporter.Sync(ctx, rejected, alert.Input{
Key: key, Severity: alert.SeverityAction, Condition: true,
Title: fmt.Sprintf("The credential bound to %q is not working", service),
Body: "Secret unresolved; runs cannot authenticate. Check the binding.",
Link: "/settings/credential-bindings",
})
}()
}
+82 -94
View File
@@ -32,6 +32,7 @@ import (
"github.com/samcharles93/archie-core/internal/daemon"
"github.com/samcharles93/archie-core/internal/domain/access"
"github.com/samcharles93/archie-core/internal/domain/agent"
"github.com/samcharles93/archie-core/internal/domain/alert"
"github.com/samcharles93/archie-core/internal/domain/applystatus"
"github.com/samcharles93/archie-core/internal/domain/health"
"github.com/samcharles93/archie-core/internal/domain/identity"
@@ -48,6 +49,8 @@ import (
"github.com/samcharles93/archie-core/internal/infrastructure/configuration"
"github.com/samcharles93/archie-core/internal/infrastructure/eventbus/nats"
"github.com/samcharles93/archie-core/internal/infrastructure/extension"
"github.com/samcharles93/archie-core/internal/infrastructure/kitrun"
"github.com/samcharles93/archie-core/internal/infrastructure/logrpc"
"github.com/samcharles93/archie-core/internal/infrastructure/messagingrpc"
"github.com/samcharles93/archie-core/internal/infrastructure/modelcatalog"
"github.com/samcharles93/archie-core/internal/infrastructure/staterpc"
@@ -64,9 +67,10 @@ import (
// reverse at shutdown, matching the LIFO ordering of the deferred calls
// they replace.
type boot struct {
catalog *modelcatalog.Catalog
cfg config.Config
log *slog.Logger
catalog *modelcatalog.Catalog
cfg config.Config
log *slog.Logger
playbooks *controlplane.LivePlaybooks
// stderrLog keeps setupLogging on stderr, never touching cfg.Log.File. An
// offline command reads the bootstrap config but is not the daemon: creating
@@ -94,6 +98,7 @@ type boot struct {
// [services.state].target. It is set by openStateStoreAdapter, which
// requires [services.state].target to be set.
stateStore storecontract.TaskStore
taskModels daemon.TaskModels
// accessChain is the daemon's policy engine, built once from the stored
// policies (openAccessChain); accessProblems is what the readiness
// surface reports. Both are nil when no policy store is wired.
@@ -103,7 +108,6 @@ type boot struct {
// (daemon.RunCredentialIssuer), wrapping the same *staterpc.Client as
// stateStore, which also verifies a credential for a worktree push.
stateStoreGrants *staterpc.GrantIssuer
stateStoreToken string
controlPlane *controlplane.Client
// controlPlaneRPC is the State Store's control-plane transport, kept so the
// daemon root can build its workflow-definitions client from it
@@ -158,8 +162,14 @@ type boot struct {
containerPool *container.Pool
// kitLauncher starts Kit profile tasks; nil when the egress path cannot
// be built on this host.
kitLauncher daemon.KitLauncher
storeBackend storage.Backend
kitLauncher *kitrun.Launcher
// modelEgress gives native tasks model access through the egress proxy;
// nil when the proxy cannot start.
modelEgress daemon.ModelEgress
// sessionGrants records live setup sessions' granted credential services
// for the egress proxy's resolver.
sessionGrants *sessionGrants
storeBackend storage.Backend
// runtimeVersions records the version of every control-plane kind boot's
// layering applied. Set once at boot before the runtime-resource watches
@@ -178,14 +188,6 @@ type boot struct {
// schedulingEngine is the cron/scheduling ticker engine (setupScheduling).
schedulingEngine *scheduling.Engine
// kindWorkflows/labelWorkflows are the resolved kind/label -> workflow
// routing bindings loaded by loadWorkflowRouting. They are handed to the
// daemon so runViaAgent can carry them in taskrun.Request: workflow.Route
// runs in the archie-agent process, which never sees the daemon's
// package-level workflow state.
kindWorkflows workflow.KindWorkflows
labelWorkflows workflow.LabelWorkflows
// agentStatus records the most recent version/install-type an
// archie-agent worker reported about itself. Allocated up front, before
// any gateway goroutine (Telegram, webui) that might read it via
@@ -200,6 +202,13 @@ type boot struct {
// signal path rather than inventing a new one.
shutdown context.CancelFunc
// alertChecks re-evaluates the configuration alerts when the configuration
// changes. Nil until the daemon's run loop starts them.
alertChecks *alertChecks
// credentialWatch reports bound credentials that stop resolving. Nil
// when the State Store has no alert surface.
credentialWatch *credentialWatch
cleanups []func()
}
@@ -263,17 +272,20 @@ func (b *boot) openStateStoreAdapter(ctx context.Context) error {
if target == "" {
return fmt.Errorf("services.state.target is required: archied/archie-gateway no longer own archie.db; the standalone archie-state-store process owns it (docs/prds/state-store-contract.md §12 step 7)")
}
client, cleanup, err := servicekit.StateStoreClient(presence.Daemon, b.cfg.Services, b.secrets)
client, cleanup, err := servicekit.StateStoreClient(presence.Daemon, b.cfg)
if err != nil {
b.log.Error("state store adapter", "err", err)
return err
}
b.stateStore = client
if alerts, ok := b.stateStore.(alert.Store); ok {
b.credentialWatch = &credentialWatch{reporter: alert.Reporter{Raiser: alerts, Log: b.log, By: "archied"}}
}
b.taskModels = client
b.controlPlaneRPC = client.ControlPlane()
b.controlPlane = controlplane.NewRPCClient(b.controlPlaneRPC)
b.applyStatus = applystatus.New(b.processName, client, b.log)
b.stateStoreGrants = &staterpc.GrantIssuer{Client: client}
b.stateStoreToken = b.cfg.Services.ResolvedToken(config.ServiceNameState, b.secrets.Getenv)
b.addCleanup(cleanup)
b.startSecretEngines(ctx, client)
return nil
@@ -340,7 +352,7 @@ func (b *boot) setupObservability(ctx context.Context) {
const reactionStreamMaxAge = 24 * time.Hour
func (b *boot) connectNATS(ctx context.Context) error { //nolint:nestif // embedded broker discovery and startup require explicit fallback branches
func (b *boot) connectNATS(ctx context.Context) error { // embedded broker discovery and startup require explicit fallback branches
cfg, log := b.cfg, b.log
url := cfg.NATS.URL
var natsToken string
@@ -462,70 +474,25 @@ func (b *boot) setupContainers(ctx context.Context) func() {
// setupGatewayClient dials the Gateway, which owns every model call. The daemon
// only enqueues the chat tasks scheduled workflows create.
func (b *boot) setupGatewayClient() error {
func (b *boot) setupGatewayClient(ctx context.Context) error {
b.setupChatTasks(b.cfg)
contract, cleanup, err := composeChatContract(b.cfg.Services, b.secrets)
contract, cleanup, err := composeChatContract(b.cfg)
if err != nil {
return err
}
b.addCleanup(cleanup)
b.chat = &webui.ChatService{Contract: contract}
messaging, closeMessaging, err := composeMessaging(b.cfg.Services, b.secrets)
messaging, closeMessaging, err := composeMessaging(b.cfg)
if err != nil {
return err
}
b.addCleanup(closeMessaging)
b.messaging = messaging
go postOutcomes(ctx, b.stateStore, messaging.Deliver, b.log)
b.setupReadinessProbes()
return nil
}
func (b *boot) loadWorkflows() error {
cfg, log := b.cfg, b.log
if err := b.loadWorkflowRouting(cfg, log); err != nil {
return err
}
log.Info("workflow step registry built", "shipped_workflows", len(workflow.ShippedDefinitions().Definitions))
return nil
}
func (b *boot) loadWorkflowRouting(cfg config.Config, log *slog.Logger) error {
kindWorkflows, err := workflow.LoadKindWorkflowsYAML(cfg.WorkflowRoutingFile)
if err != nil {
log.Error("workflow routing file load failed", "path", cfg.WorkflowRoutingFile, "err", err)
return err
}
labelWorkflows, err := workflow.LoadLabelWorkflowsYAML(cfg.WorkflowLabelsFile)
if err != nil {
log.Error("workflow labels file load failed", "path", cfg.WorkflowLabelsFile, "err", err)
return err
}
dirKindWorkflows, dirLabelWorkflows, err := workflow.LoadPlaybookDirs(cfg.PlaybookDirs)
if err != nil {
log.Error("playbook dirs load failed", "dirs", cfg.PlaybookDirs, "err", err)
return err
}
kindWorkflows, err = workflow.MergeKindWorkflows(kindWorkflows, dirKindWorkflows)
if err != nil {
log.Error("workflow binding collision between routing file and playbook dir", "err", err)
return err
}
labelWorkflows, err = workflow.MergeLabelWorkflows(labelWorkflows, dirLabelWorkflows)
if err != nil {
log.Error("workflow binding collision between labels file and playbook dir", "err", err)
return err
}
workflow.SetKindWorkflows(kindWorkflows)
workflow.SetLabelWorkflows(labelWorkflows)
b.kindWorkflows = kindWorkflows
b.labelWorkflows = labelWorkflows
return nil
}
func (b *boot) buildWorktreeManager() {
cfg := b.cfg
b.trees = &worktree.Manager{
@@ -571,13 +538,18 @@ func (b *boot) registerNATSRPC() error {
log.Error("nats connection unavailable for task RPC", "err", err)
return err
}
unsubscribe, err := registerTaskRPCServers(coreConn, b.forgeClient, b.trees, b.identityRunners, b.stateStoreGrants.Client, log)
unsubscribe, err := registerTaskRPCServers(coreConn, b.forgeClient, b.trees, b.identityRunners, b.stateStoreGrants.Client, installedAuthorities{b.stateStoreGrants.Client}, log)
if err != nil {
log.Error("task RPC server registration failed", "err", err)
return err
}
b.addCleanup(unsubscribe)
stopLogs, err := logrpc.RegisterDaemon(coreConn, b.logFeed, log)
if err != nil {
return fmt.Errorf("register daemon logs: %w", err)
}
b.addCleanup(stopLogs)
unsubscribeSystemLogs, err := subscribeSystemLogs(coreConn, b.taskLogs, log)
if err != nil {
log.Error("system log subscribe failed", "err", err)
@@ -606,11 +578,12 @@ func (b *boot) buildDaemon() {
b.d = &daemon.Daemon{
Cfg: b.cfgHolder,
ConnectedNATS: daemon.NATSEndpoint{URL: b.natsURL, Token: b.natsToken, Scoped: b.cfg.NATS.Mode != config.NATSModeExternal},
ConnectedStateStore: daemon.StateStoreEndpoint{URL: strings.TrimSpace(b.cfg.Services.Get(config.ServiceNameState).Target), Token: b.stateStoreToken},
ConnectedStateStore: daemon.StateStoreEndpoint{URL: strings.TrimSpace(b.cfg.Services.Get(config.ServiceNameState).Target), CA: b.stateStoreCA()},
Store: b.stateStore,
Bus: b.bus,
Forge: b.forgeClient,
Trees: b.trees,
PackageAuthorities: installedAuthorities{b.stateStoreGrants.Client},
Storage: b.storeBackend,
Log: log,
Tasks: b.natsClient,
@@ -618,15 +591,15 @@ func (b *boot) buildDaemon() {
RunCredentials: b.stateStoreGrants,
ContainerPool: b.containerPool,
KitLauncher: b.kitLauncher,
ModelEgress: b.modelEgress,
TaskModels: b.taskModels,
Identities: b.identityRunners,
RootIdentityID: identity.StableID(servicekit.IdentityNames(b.cfg)[0]),
TaskLogs: b.taskLogs,
AgentStatus: b.agentStatus,
KindWorkflows: b.kindWorkflows,
LabelWorkflows: b.labelWorkflows,
WorkflowDefinitions: b.workflowDefinitions,
WorkflowEnablement: b.controlPlane,
Playbooks: controlplane.NewLivePlaybooks(b.controlPlaneRPC, b.log),
Playbooks: b.livePlaybooks(),
}
if identities, ok := b.stateStore.(identity.Repository); ok {
@@ -717,29 +690,16 @@ func (b *boot) publishConfig(ctx context.Context, cfg config.Config) {
// The daemon owns the published configuration view.
if b.processName != applystatus.Gateway {
b.publishConfigSnapshot(ctx)
b.recheckAlerts(ctx)
}
}
func (b *boot) configOrigins() []webui.ConfigOrigin {
provenance := b.currentProvenance.Load()
if provenance == nil {
return nil
}
origins := make([]webui.ConfigOrigin, 0, len(provenance.Origins))
for _, origin := range provenance.Origins {
origins = append(origins, webui.ConfigOrigin{
Path: origin.Path, Role: string(origin.Role), Layer: string(origin.Layer), Feature: string(origin.Feature),
})
}
return origins
}
func (b *boot) configViewInput() webui.ConfigViewInput {
catalog, _ := b.catalogState()
catalog := b.catalogState()
in := webui.ConfigViewInput{
Config: b.cfgHolder.Get(),
Provenance: b.configOrigins(),
Catalog: catalogView(catalog),
Config: b.cfgHolder.Get(),
Catalog: catalogView(catalog),
SecretEngines: b.secrets.List(),
}
if b.lastReload != nil {
status := b.lastReload()
@@ -805,12 +765,14 @@ func (b *boot) startServices(ctx context.Context) error {
func (b *boot) setupBackends(ctx context.Context) error {
closeContainers := b.setupContainers(ctx)
// Registered before the Kit launcher so its cleanups (the harness
// listener and the egress relay) run first: cleanups unwind LIFO, and a
// session teardown needs the Docker client the pool still owns.
b.addCleanup(closeContainers)
err := b.connectNATS(ctx)
if err == nil {
b.setupKitLauncher(ctx)
}
b.addCleanup(closeContainers)
return err
}
@@ -826,11 +788,16 @@ func (b *boot) runLoop(ctx context.Context, once bool) error {
b.announceReady()
if once {
b.d.Cycle(ctx)
// A drain pass no longer waits for the tasks it submitted, so a
// single-cycle invocation waits for them here: exiting while their
// containers still run would kill work this pass accepted.
b.d.WaitForTasks()
return nil
}
b.log.Info("archied running", "repos", len(b.cfg.Repos), "poll", b.cfg.PollInterval.Std().String(), "label", b.cfg.Label)
b.startDrainMonitor(ctx)
b.startWatchdog(ctx)
b.startAlerts(ctx)
if err := b.d.Run(ctx); err != nil && ctx.Err() == nil {
b.log.Error("daemon exited", "err", err)
return err
@@ -854,7 +821,7 @@ func (b *boot) addCleanup(fn func()) {
b.cleanups = append(b.cleanups, fn)
}
func (b *boot) catalogState() (modelcatalog.Snapshot, []string) { return b.catalog.State() }
func (b *boot) catalogState() modelcatalog.Snapshot { return b.catalog.State() }
// loadCatalog layers the model catalog under the file config. A catalog that
// cannot be read leaves the configured providers and models in effect.
@@ -866,7 +833,7 @@ func (b *boot) loadCatalog(ctx context.Context) {
}
models := modelcatalog.Apply(&b.cfg, snapshot)
b.cfgHolder.Set(b.cfg.Clone())
b.catalog.Set(snapshot, models)
b.catalog.Set(snapshot)
b.log.Info("model catalog loaded", "providers", len(snapshot.Providers), "models", len(models))
}
@@ -881,7 +848,7 @@ func (b *boot) refreshModelCatalog(ctx context.Context) error {
if err != nil {
return err
}
b.catalog.Set(snapshot, models)
b.catalog.Set(snapshot)
b.publishConfig(ctx, cfg)
b.log.Info("model catalog refreshed", "providers", len(snapshot.Providers), "models", len(models))
return nil
@@ -894,3 +861,24 @@ func (b *boot) startModelCatalogRefresh(ctx context.Context) {
}
})
}
func (b *boot) livePlaybooks() *controlplane.LivePlaybooks {
b.playbooks = controlplane.NewLivePlaybooks(b.controlPlaneRPC, b.log)
b.playbooks.ApplyStatus = b.applyStatus
return b.playbooks
}
// playbookRefreshInterval is how often archied re-reads the resources it
// would otherwise load only on first use, so their apply status stays reported.
const playbookRefreshInterval = 30 * time.Second
// stateStoreCA is the CA a task container verifies the State Store with;
// nil without [tls].
func (b *boot) stateStoreCA() []byte {
id, err := servicekit.Identity(b.cfg.TLS, presence.Daemon)
if err != nil {
b.log.Warn("state store CA for containers", "err", err)
return nil
}
return id.CAPEM()
}
+17 -11
View File
@@ -3,27 +3,29 @@ package archied
import (
"google.golang.org/grpc"
"github.com/samcharles93/archie-core/internal/app/servicekit"
"github.com/samcharles93/archie-core/internal/config"
"github.com/samcharles93/archie-core/internal/domain/presence"
"github.com/samcharles93/archie-core/internal/gateway"
"github.com/samcharles93/archie-core/internal/infrastructure/gatewayrpc"
"github.com/samcharles93/archie-core/internal/infrastructure/messagingrpc"
"github.com/samcharles93/archie-core/internal/secret"
)
// composeChatContract dials the standalone archie-gateway, the sole owner of
// ChatContract, from the daemon's own configuration key. Token resolution is
// the daemon's (it owns the config keys and the secret registry); the dial
// itself and the fail-closed non-loopback rule belong to the transport, so
// they come from gatewayrpc.Dial.
func composeChatContract(services config.Services, secrets *secret.Registry, options ...grpc.DialOption) (gateway.ChatContract, func(), error) {
target, err := services.RequireTarget(config.ServiceNameGateway)
// ChatContract, from the daemon's own configuration key.
func composeChatContract(cfg config.Config, options ...grpc.DialOption) (gateway.ChatContract, func(), error) {
target, err := cfg.Services.RequireTarget(config.ServiceNameGateway)
if err != nil {
return nil, nil, err
}
id, err := servicekit.Identity(cfg.TLS, presence.Daemon)
if err != nil {
return nil, nil, err
}
// Assigned through a typed variable rather than returned directly: a
// direct forward would hand a failed dial's nil *Client back as a
// non-nil ChatContract interface.
client, cleanup, err := gatewayrpc.Dial(target, services.ResolvedToken(config.ServiceNameGateway, secrets.Getenv), options...)
client, cleanup, err := gatewayrpc.Dial(target, id, options...)
if err != nil {
return nil, nil, err
}
@@ -32,10 +34,14 @@ func composeChatContract(services config.Services, secrets *secret.Registry, opt
// composeMessaging builds the Messaging Service client. The connection is
// lazy, so an absent Messaging Service fails each delivery, not the daemon.
func composeMessaging(services config.Services, secrets *secret.Registry) (*messagingrpc.Client, func(), error) {
target, err := services.RequireTarget(config.ServiceNameMessaging)
func composeMessaging(cfg config.Config) (*messagingrpc.Client, func(), error) {
target, err := cfg.Services.RequireTarget(config.ServiceNameMessaging)
if err != nil {
return nil, nil, err
}
return messagingrpc.Dial(target, services.ResolvedToken(config.ServiceNameMessaging, secrets.Getenv))
id, err := servicekit.Identity(cfg.TLS, presence.Daemon)
if err != nil {
return nil, nil, err
}
return messagingrpc.Dial(target, id)
}
+9 -9
View File
@@ -24,7 +24,7 @@ func (b *boot) runtimeConfig(ctx context.Context, base config.Config) (config.Co
if settings := b.executionSettings.Load(); settings != nil {
applyExecutionBudgets(cfg, *settings)
}
})
}, nil)
}
func (b *boot) loadRuntimeConfig(ctx context.Context) error {
@@ -52,7 +52,7 @@ func (b *boot) reloadConfig(ctx context.Context, doc *configuration.Document) er
// drop them from the running config even though the file is
// unchanged. Re-apply the same merge (idempotent: a catalog that
// failed to load merges to identity).
catalog, _ := b.catalogState()
catalog := b.catalogState()
modelcatalog.Apply(&doc.Config, catalog)
// Bounded: this runs on the signal loop, which handles nothing else
// while it waits, and a State Store that has stopped answering must
@@ -63,15 +63,17 @@ func (b *boot) reloadConfig(ctx context.Context, doc *configuration.Document) er
if err != nil {
return fmt.Errorf("runtime settings unavailable: %w", err)
}
doc.Config = cfg
old := b.cfgHolder.Get()
restartFields := changedNonReloadableFields(old, cfg)
cfg = preserveForgeIdentity(old, cfg)
doc.Config = cfg
b.currentProvenance.Store(&doc.Provenance)
b.publishConfig(ctx, cfg)
// The container pool reads the published config on every acquire, but the
// dispatcher holds its limit; a reloaded containers.max_concurrency must
// reach it the same way a control-plane update does.
b.resizeTaskDispatcher(cfg.Containers.MaxConcurrency)
if fields := changedNonReloadableFields(old, cfg); len(fields) > 0 {
if fields := restartFields; len(fields) > 0 {
b.log.Warn("config reloaded; some changes require a restart",
"fields", fields, "paths", doc.Provenance.Paths())
} else {
@@ -210,9 +212,6 @@ func applyExecutionBudgets(cfg *config.Config, settings workflow.ExecutionSettin
MaxSteps: settings.MaxModelToolSteps, WallClock: config.Duration(settings.MaxRuntime), GateMaxFailures: settings.MaxConsecutiveGateFailures,
TaskWallClock: config.Duration(settings.MaxTaskRuntime),
}
for i := range cfg.Identities {
cfg.Identities[i].Budgets = cfg.Budgets
}
}
// startLiveSettings starts everything a live control-plane update needs
@@ -263,8 +262,9 @@ func (b *boot) startLiveSettings(ctx context.Context) error {
// restart-required. Each kind is watched in its own goroutine, exactly the
// shape the workflow-execution-settings watch established.
var runtimeResourceKinds = []string{
controlplane.TaskPolicyKind,
controlplane.ProviderSettingsKind,
controlplane.ModelRoleAssignmentsKind,
controlplane.ModelAliasesKind,
controlplane.RepositoryPoliciesKind,
controlplane.SchedulingPolicyKind,
controlplane.ToolSettingsKind,
@@ -370,7 +370,7 @@ func (b *boot) applyRuntimeResourceUpdate(ctx context.Context, kind string, upda
// this read returns can lack the catalog's model limits and discovered
// providers. The merge is idempotent, and reloadConfig re-applies it
// before layering for the same reason.
catalog, _ := b.catalogState()
catalog := b.catalogState()
modelcatalog.Apply(&base, catalog)
cfg, _, err := b.runtimeConfig(ctx, base)
if err != nil {
+251
View File
@@ -0,0 +1,251 @@
package archied
import (
"context"
"errors"
"io"
"log/slog"
"net"
"slices"
"sync"
"time"
"google.golang.org/grpc"
"google.golang.org/grpc/codes"
"google.golang.org/grpc/status"
"github.com/samcharles93/archie-core/internal/app/servicekit"
"github.com/samcharles93/archie-core/internal/config"
"github.com/samcharles93/archie-core/internal/container"
pb "github.com/samcharles93/archie-core/internal/contracts/harness/v1"
"github.com/samcharles93/archie-core/internal/domain/presence"
"github.com/samcharles93/archie-core/internal/infrastructure/harnesssession"
"github.com/samcharles93/archie-core/internal/infrastructure/kitrun"
)
// setupSessionTTL bounds a setup container's life independently of the pool's
// max-uptime reaper: a login that never finishes must not hold a container
// until the reaper's (longer) cap.
const setupSessionTTL = 30 * time.Minute
// sessionGrants records the credential services each live setup session is
// bound for. A session's proxy token names no task, so the egress proxy's run
// resolver consults this registry: without it, the token endpoint is never
// intercepted and no OAuth token is ever captured.
type sessionGrants struct {
mu sync.RWMutex
m map[string]sessionGrant
}
type sessionGrant struct {
org string
services []string
}
func (g *sessionGrants) Grant(token, org string, services []string) {
g.mu.Lock()
defer g.mu.Unlock()
if g.m == nil {
g.m = map[string]sessionGrant{}
}
g.m[token] = sessionGrant{org: org, services: slices.Clone(services)}
}
func (g *sessionGrants) Revoke(token string) {
g.mu.Lock()
defer g.mu.Unlock()
delete(g.m, token)
}
// GrantedFor reports the org and granted services a session token carries.
func (g *sessionGrants) GrantedFor(token string) (string, []string, bool) {
g.mu.RLock()
defer g.mu.RUnlock()
grant, ok := g.m[token]
return grant.org, grant.services, ok
}
// harnessServer serves the daemon's harness session contract.
type harnessServer struct {
pb.UnimplementedHarnessServiceServer
sessions *harnesssession.Manager
// runs serves the run target; nil answers Unavailable.
runs *runView
log *slog.Logger
}
// Open dispatches on the first frame's target. Setup sessions open a PTY; a run
// target is a read-only view of the task's running sandbox.
func (h *harnessServer) Open(stream pb.HarnessService_OpenServer) error {
first, err := stream.Recv()
if err != nil {
return err
}
target := first.GetTarget()
if target == nil || target.GetOrg() == "" {
return status.Error(codes.InvalidArgument, "the first frame must name a session target with an org")
}
switch t := target.GetTarget().(type) {
case *pb.OpenTarget_Setup:
if t.Setup.GetProfile() == "" {
return status.Error(codes.InvalidArgument, "the setup target names no profile")
}
return h.openSetup(stream, target.GetOrg(), t.Setup.GetProfile(), first.GetSize())
case *pb.OpenTarget_Run:
if t.Run.GetTaskId() <= 0 {
return status.Error(codes.InvalidArgument, "the run target names no task")
}
return h.openRun(stream, target.GetOrg(), t.Run)
default:
return status.Error(codes.InvalidArgument, "the target names no session")
}
}
// openSetup opens one setup session and pumps frames until the shell exits or
// either side closes. Teardown runs on every return.
func (h *harnessServer) openSetup(stream pb.HarnessService_OpenServer, org, profile string, size *pb.ConsoleSize) error {
session, err := h.sessions.Open(stream.Context(), org, profile, int(size.GetRows()), int(size.GetCols()))
if err != nil {
h.log.Warn("setup session failed to open", "org", org, "profile", profile, "err", err)
// The failure rides the stream as a frame, not only as a status: the
// dashboard's terminal shows the frame's text, which is the operator's
// only honest account of why the session did not open.
return stream.Send(&pb.OpenResponse{Frame: &pb.OpenResponse_Error{Error: err.Error()}})
}
defer func() {
_ = session.Close()
h.log.Info("setup session closed", "org", org, "profile", profile)
}()
h.log.Info("setup session opened", "org", org, "profile", profile)
output := make(chan error, 1)
go func() { output <- h.pumpOutput(stream, session) }()
input := make(chan error, 1)
go func() { input <- h.pumpInput(stream, session) }()
select {
case err := <-output:
if err != nil {
return status.Error(codes.Internal, err.Error())
}
code, codeErr := session.ExitCode(context.WithoutCancel(stream.Context()))
if codeErr != nil {
return nil
}
return stream.Send(&pb.OpenResponse{Frame: &pb.OpenResponse_ExitCode{ExitCode: int32(code)}})
case err := <-input:
if errors.Is(err, io.EOF) {
return nil
}
return err
}
}
// pumpOutput copies terminal output to the stream until the shell ends.
func (h *harnessServer) pumpOutput(stream pb.HarnessService_OpenServer, session *harnesssession.Session) error {
buf := make([]byte, 32<<10)
for {
n, err := session.Read(buf)
if n > 0 {
frame := &pb.OpenResponse{Frame: &pb.OpenResponse_Stdout{Stdout: slices.Clone(buf[:n])}}
if sendErr := stream.Send(frame); sendErr != nil {
return sendErr
}
}
if err != nil {
if errors.Is(err, io.EOF) {
return nil
}
return err
}
}
}
// pumpInput copies client frames to the session until the client closes.
func (h *harnessServer) pumpInput(stream pb.HarnessService_OpenServer, session *harnesssession.Session) error {
for {
req, err := stream.Recv()
if err != nil {
return err
}
if stdin, ok := req.GetFrame().(*pb.OpenRequest_Stdin); ok {
if _, err := session.Write(stdin.Stdin); err != nil {
return err
}
}
}
}
// setupHarness serves the harness session contract on the configured listen
// address. It is skipped, with the reason logged, when no service token is
// configured: a session is a shell in a container that holds an org's
// credential, so the token is required even on loopback.
func (b *boot) setupHarness(ctx context.Context) {
conn := b.cfg.Services.Get(config.ServiceNameHarness)
listen, err := servicekit.ResolveListen(config.ServiceNameHarness, "", conn.Listen)
if err != nil {
b.log.Warn("harness sessions disabled", "err", err)
return
}
id, err := servicekit.Identity(b.cfg.TLS, presence.Daemon)
if err != nil {
b.log.Warn("harness sessions disabled", "err", err)
return
}
if id == nil {
b.log.Warn("harness sessions disabled: a session is a shell holding an org's credential, so it needs [tls] ca_file and cert_dir")
return
}
creds, err := id.ServerOption(listen, false)
if err != nil {
b.log.Warn("harness sessions disabled", "err", err)
return
}
ln, err := (&net.ListenConfig{}).Listen(ctx, "tcp", listen)
if err != nil {
b.log.Warn("harness sessions disabled: listen", "err", err)
return
}
manager := &harnesssession.Manager{
Launcher: b.kitLauncher,
Client: b.containerPool.Client(),
Profile: func(name string) (config.AgentProfile, error) { return b.cfgHolder.Get().Containers.Profile(name) },
TTL: setupSessionTTL,
}
srv := grpc.NewServer(append(servicekit.ServerLimits(), creds)...)
pb.RegisterHarnessServiceServer(srv, &harnessServer{sessions: manager, runs: b.runView(), log: b.log})
go func() {
if err := srv.Serve(ln); err != nil && !errors.Is(err, grpc.ErrServerStopped) {
b.log.Error("harness listener stopped", "err", err)
}
}()
b.addCleanup(func() {
// An open session stream would hold GracefulStop until its TTL; give
// in-flight frames a moment, then stop. Mirrors the State Store's
// shutdown so a session cannot hold shutdown to the watchdog leash.
timer := time.AfterFunc(10*time.Second, srv.Stop)
srv.GracefulStop()
timer.Stop()
_ = ln.Close()
})
b.log.Info("harness sessions enabled", "addr", ln.Addr().String())
}
var _ kitrun.GrantRegistry = (*sessionGrants)(nil)
// runView wires the run target to the task store and the container pool. The
// store is the daemon's own contract client, so a task it cannot read answers
// Unavailable rather than guessing.
func (b *boot) runView() *runView {
if b.stateStore == nil || b.containerPool == nil {
return nil
}
cli := b.containerPool.Client()
return &runView{
tasks: b.stateStore,
follow: func(ctx context.Context, taskID, attempt int64) (io.ReadCloser, error) {
return container.FollowLogs(ctx, cli, kitrun.TaskContainer(taskID, attempt), runTail)
},
}
}
+97
View File
@@ -0,0 +1,97 @@
package archied
import (
"context"
"errors"
"fmt"
"io"
"github.com/containerd/errdefs"
"google.golang.org/grpc/codes"
"google.golang.org/grpc/status"
"github.com/samcharles93/archie-core/internal/container"
pb "github.com/samcharles93/archie-core/internal/contracts/harness/v1"
"github.com/samcharles93/archie-core/internal/domain/org"
"github.com/samcharles93/archie-core/internal/domain/workflow"
)
// runTail is how much recent output a view opens with, in lines.
const runTail = 200
// taskLookup reads the task a run target names.
type taskLookup interface {
TaskByID(ctx context.Context, id int64) (*workflow.Task, error)
}
// runView resolves a run target to its container and follows its output. It
// is read-only by construction: it holds no exec and no stdin path, so a view
// confers no authority over the container.
type runView struct {
tasks taskLookup
follow func(ctx context.Context, taskID, attempt int64) (io.ReadCloser, error)
}
// open refuses a task the daemon does not know, one that is not running, and
// an attempt that is not the running one. Each refusal states its reason: the
// container is gone with the attempt, so there is nothing to answer
// "unavailable" about.
func (v *runView) open(ctx context.Context, target *pb.RunTarget) (io.ReadCloser, error) {
t, err := v.tasks.TaskByID(ctx, target.GetTaskId())
if err != nil {
return nil, status.Error(codes.Unavailable, "the task store is unavailable: "+err.Error())
}
switch {
case t == nil || t.OrgOrDefault() != string(org.OrgFromContext(ctx)):
return nil, status.Errorf(codes.NotFound, "task %d is unknown", target.GetTaskId())
case t.Status != workflow.StatusRunning:
return nil, status.Errorf(codes.NotFound, "task %d is %s, not running; its sandbox is gone", t.ID, t.Status)
case int64(t.Attempt) != target.GetAttempt():
return nil, status.Errorf(codes.NotFound, "attempt %d is not running; task %d is on attempt %d", target.GetAttempt(), t.ID, t.Attempt)
}
logs, err := v.follow(ctx, t.ID, target.GetAttempt())
if err != nil {
if errdefs.IsNotFound(err) || errors.Is(err, container.ErrNotRunning) {
return nil, status.Errorf(codes.NotFound, "task %d has no running sandbox: %v", t.ID, err)
}
return nil, status.Errorf(codes.Unavailable, "the sandbox view is unavailable: %v", err)
}
return logs, nil
}
// openRun streams a running task's sandbox output until the task's container
// ends or the client leaves. Client frames are ignored: this is the peek, not
// a takeover.
func (h *harnessServer) openRun(stream pb.HarnessService_OpenServer, orgID string, target *pb.RunTarget) error {
if h.runs == nil {
return status.Error(codes.Unavailable, "run-scoped views need the task store and the container pool")
}
logs, err := h.runs.open(org.WithOrg(stream.Context(), org.OrgID(orgID)), target)
if err != nil {
return err
}
defer func() { _ = logs.Close() }()
// An empty frame tells the client the view opened before the sandbox has
// printed anything.
if err := stream.Send(&pb.OpenResponse{Frame: &pb.OpenResponse_Stdout{}}); err != nil {
return err
}
h.log.Info("run view opened", "task", target.GetTaskId(), "attempt", target.GetAttempt())
defer h.log.Info("run view closed", "task", target.GetTaskId(), "attempt", target.GetAttempt())
buf := make([]byte, 32<<10)
for {
n, err := logs.Read(buf)
if n > 0 {
if sendErr := stream.Send(&pb.OpenResponse{Frame: &pb.OpenResponse_Stdout{Stdout: append([]byte(nil), buf[:n]...)}}); sendErr != nil {
return sendErr
}
}
if err != nil {
if err == io.EOF || stream.Context().Err() != nil {
return nil
}
return status.Error(codes.Unavailable, fmt.Sprintf("the sandbox output ended: %v", err))
}
}
}
+65
View File
@@ -0,0 +1,65 @@
package archied
import (
"context"
"errors"
"io"
"strings"
"testing"
"github.com/containerd/errdefs"
"google.golang.org/grpc/codes"
"google.golang.org/grpc/status"
pb "github.com/samcharles93/archie-core/internal/contracts/harness/v1"
"github.com/samcharles93/archie-core/internal/domain/org"
"github.com/samcharles93/archie-core/internal/domain/workflow"
)
type fakeTasks struct {
task *workflow.Task
err error
}
func (f fakeTasks) TaskByID(context.Context, int64) (*workflow.Task, error) { return f.task, f.err }
func TestRunViewOpen(t *testing.T) {
running := &workflow.Task{ID: 7, Status: workflow.StatusRunning, Attempt: 2}
cases := []struct {
name string
tasks fakeTasks
attempt int64
follow error
code codes.Code
reason string
}{
{"running attempt", fakeTasks{task: running}, 2, nil, codes.OK, ""},
{"unknown task", fakeTasks{}, 2, nil, codes.NotFound, "unknown"},
{"finished task", fakeTasks{task: &workflow.Task{ID: 7, Status: workflow.StatusParked, Attempt: 2}}, 2, nil, codes.NotFound, "not running"},
{"earlier attempt", fakeTasks{task: running}, 1, nil, codes.NotFound, "attempt 1 is not running"},
{"container gone", fakeTasks{task: running}, 2, errdefs.ErrNotFound, codes.NotFound, "no running sandbox"},
{"store down", fakeTasks{err: errors.New("dial")}, 2, nil, codes.Unavailable, "unavailable"},
{"other org", fakeTasks{task: &workflow.Task{ID: 7, Org: "other", Status: workflow.StatusRunning, Attempt: 2}}, 2, nil, codes.NotFound, "unknown"},
{"docker down", fakeTasks{task: running}, 2, errors.New("connection refused"), codes.Unavailable, "unavailable"},
}
for _, tc := range cases {
t.Run(tc.name, func(t *testing.T) {
v := &runView{tasks: tc.tasks, follow: func(context.Context, int64, int64) (io.ReadCloser, error) {
if tc.follow != nil {
return nil, tc.follow
}
return io.NopCloser(strings.NewReader("")), nil
}}
rc, err := v.open(org.WithOrg(t.Context(), org.DefaultOrgID), &pb.RunTarget{TaskId: 7, Attempt: tc.attempt})
if got := status.Code(err); got != tc.code {
t.Fatalf("code = %v (%v), want %v", got, err, tc.code)
}
if err != nil && !strings.Contains(err.Error(), tc.reason) {
t.Fatalf("reason %q does not state %q", err, tc.reason)
}
if rc != nil {
_ = rc.Close()
}
})
}
}
@@ -0,0 +1,21 @@
package archied
import (
"context"
"github.com/samcharles93/archie-core/internal/domain/storepkg"
)
// installedAuthorities answers a package's accepted authority from the State
// Store. A package the operator has not accepted has none.
type installedAuthorities struct {
packages storepkg.Installations
}
func (a installedAuthorities) AcceptedAuthority(ctx context.Context, name string) (storepkg.Authority, error) {
installed, err := a.packages.GetInstalled(ctx, name)
if err != nil || installed.AcceptedAuthority == nil {
return storepkg.Authority{}, err
}
return *installed.AcceptedAuthority, nil
}
+17 -8
View File
@@ -42,11 +42,6 @@ func (b *boot) setupKitLauncher(ctx context.Context) {
log.Warn("kit harness runs disabled: egress CA", "err", err)
return
}
fetcher, err := fetch.New(fetch.WithDockerCredentials())
if err != nil {
log.Warn("kit harness runs disabled: kit registry client", "err", err)
return
}
ln, err := (&net.ListenConfig{}).Listen(ctx, "tcp", net.JoinHostPort(pool.HostGateway(), "0"))
if err != nil {
log.Warn("kit harness runs disabled: egress proxy listener", "err", err)
@@ -60,8 +55,10 @@ func (b *boot) setupKitLauncher(ctx context.Context) {
return
}
// The proxy resolves each credential from the run credential the
// container logs in with, against the live grants and bindings.
resolver := runCredentialResolver{runs: b.stateStoreGrants.Client, config: b.cfgHolder, secrets: b.secrets}
// container logs in with, against the live grants and bindings; a setup
// session's token resolves through the session grants instead.
b.sessionGrants = &sessionGrants{}
resolver := runCredentialResolver{runs: b.stateStoreGrants.Client, sessions: b.sessionGrants, models: b.taskModels, config: b.cfgHolder, secrets: b.secrets, credentials: b.credentialWatch}
// Without a harness secret store, Register refuses a Kit with a
// required OAuth credential.
oauthStore, _ := b.stateStore.(egress.OAuthStore)
@@ -72,16 +69,28 @@ func (b *boot) setupKitLauncher(ctx context.Context) {
log.Error("egress proxy stopped", "err", err)
}
}()
// Native tasks need only the proxy for model access, so it serves them
// even when a Kit-only piece below fails.
b.modelEgress = modelEgress{proxy: proxy, addr: ln.Addr().String(), caFile: egress.CACertPath(caDir), config: b.cfgHolder}
fetcher, err := fetch.New(fetch.WithDockerCredentials())
if err != nil {
b.addCleanup(func() { _ = srv.Close() })
log.Warn("kit harness runs disabled: kit registry client", "err", err)
return
}
networks := egress.NewNetworks(pool.Client(), egress.RelaySpec{
Name: relayName, Image: b.cfg.Containers.Image,
Entrypoint: []string{"archie-agent"}, Cmd: cmd, Network: pool.NetworkName(),
})
b.addCleanup(shutdownEgress(networks, srv, log))
b.kitLauncher = &kitrun.Launcher{
kitLauncher := &kitrun.Launcher{
Pool: pool, Fetch: fetcher, Proxy: proxy, Networks: networks,
AgentBinary: agentBinary, CAFile: egress.CACertPath(caDir),
Config: b.cfgHolder, Secrets: b.secrets, OAuth: oauthStore,
Grants: b.sessionGrants,
}
b.kitLauncher = kitLauncher
b.setupHarness(ctx)
log.Info("kit harness runs enabled", "proxy", ln.Addr().String())
}
+28 -7
View File
@@ -8,6 +8,7 @@ import (
"context"
"crypto/sha256"
"encoding/json"
"errors"
"flag"
"fmt"
"log/slog"
@@ -29,6 +30,7 @@ import (
"github.com/samcharles93/archie-core/internal/container"
"github.com/samcharles93/archie-core/internal/daemon"
"github.com/samcharles93/archie-core/internal/domain/applystatus"
"github.com/samcharles93/archie-core/internal/domain/health"
"github.com/samcharles93/archie-core/internal/domain/presence"
"github.com/samcharles93/archie-core/internal/domain/storecontract"
"github.com/samcharles93/archie-core/internal/domain/workflow"
@@ -36,6 +38,7 @@ import (
"github.com/samcharles93/archie-core/internal/forge"
"github.com/samcharles93/archie-core/internal/forgerpc"
"github.com/samcharles93/archie-core/internal/infrastructure/configuration"
"github.com/samcharles93/archie-core/internal/infrastructure/readiness"
"github.com/samcharles93/archie-core/internal/logging"
"github.com/samcharles93/archie-core/internal/secret"
"github.com/samcharles93/archie-core/internal/storage"
@@ -123,7 +126,7 @@ func Run() int { //nolint:cyclop,funlen // the composition root's setup sequence
return 0
}
if recovering, err := resumeUpdate(context.Background()); err != nil {
if recovering, err := resumeUpdate(context.Background(), args.cfgPath); err != nil {
fmt.Fprintln(os.Stderr, err)
return 1
} else if recovering {
@@ -182,11 +185,12 @@ func Run() int { //nolint:cyclop,funlen // the composition root's setup sequence
return 1
}
if err := b.setupGatewayClient(); err != nil {
if err := b.setupGatewayClient(ctx); err != nil {
return 1
}
if ps, ok := b.stateStore.(storecontract.PresenceStore); ok {
go presence.Run(ctx, b.processName, servicekit.Build(), ps, b.healthRegistry, b.log)
b.reportBrokerPresence(ctx, ps)
}
// The daemon publishes the configuration view, which carries the catalog's
// model limits.
@@ -194,9 +198,6 @@ func Run() int { //nolint:cyclop,funlen // the composition root's setup sequence
if err := b.buildTreesAndIdentities(ctx); err != nil {
return 1
}
if err := b.loadWorkflows(); err != nil {
return 1
}
if err := b.registerNATSRPC(); err != nil {
return 1
@@ -206,6 +207,9 @@ func Run() int { //nolint:cyclop,funlen // the composition root's setup sequence
}
b.buildDaemon()
b.playbooks.Keep(ctx, playbookRefreshInterval)
b.workflowDefinitions.Keep(ctx, playbookRefreshInterval)
b.controlPlane.KeepSchedules(ctx, playbookRefreshInterval)
b.wireConfigSurfaces(ctx, args.cfgPath, args.overlayPath)
if err := b.startServices(ctx); err != nil {
@@ -288,6 +292,11 @@ func subscribeSystemLogs(nc *natsio.Conn, taskLogs *logging.TaskRegistry, log *s
}
// The NATS callback has no context of its own; the write is an append to
// an already-open sink, so there is nothing to cancel.
if entry.Fields == nil {
entry.Fields = map[string]any{}
}
entry.Fields["service"] = presence.Agent
entry.Fields["task_id"] = taskID
taskLogs.Write(context.Background(), taskID, entry)
})
if err != nil {
@@ -350,7 +359,7 @@ func subscribeAgentEvents(nc *natsio.Conn, bus *events.Bus, log *slog.Logger) (u
// tasks, and one server pair per identity answers that identity's scoped
// subjects so a container-mode task owned by a non-root identity has its RPC
// calls served by its own forge client and worktree manager.
func registerTaskRPCServers(nc *natsio.Conn, forgeClient forge.Forge, trees *worktree.Manager, identities []*daemon.IdentityRunner, runs worktreerpc.Runs, log *slog.Logger) (unsubscribe func(), err error) {
func registerTaskRPCServers(nc *natsio.Conn, forgeClient forge.Forge, trees *worktree.Manager, identities []*daemon.IdentityRunner, runs worktreerpc.Runs, authorities forgerpc.Authorities, log *slog.Logger) (unsubscribe func(), err error) {
unsubs := make([]func(), 0, 2+2*len(identities))
unsubAll := func() {
for _, u := range unsubs {
@@ -359,7 +368,7 @@ func registerTaskRPCServers(nc *natsio.Conn, forgeClient forge.Forge, trees *wor
}
registerForge := func(fg forge.Forge, identity string) error {
srv := &forgerpc.Server{Forge: fg, Runs: runs, Log: log.With("rpc_identity", identity)}
srv := &forgerpc.Server{Forge: fg, Runs: runs, Authorities: authorities, Log: log.With("rpc_identity", identity)}
u, err := srv.RegisterFor(nc, identity)
if err != nil {
return fmt.Errorf("register forgerpc%s: %w", identitySuffix(identity), err)
@@ -526,3 +535,15 @@ func resolveRegistryAuth(ref secret.SecretRef, secrets *secret.Registry, log *sl
}
return value
}
func (b *boot) reportBrokerPresence(ctx context.Context, ps storecontract.PresenceStore) {
if conn, err := b.natsClient.CoreConn(); err == nil {
registry := health.NewRegistry(readiness.NewContractProbe("connection", 0, func(context.Context) error {
if !conn.IsConnected() {
return errors.New("broker disconnected")
}
return nil
}))
go presence.Run(ctx, presence.Broker, presence.Build{Version: conn.ConnectedServerVersion(), InstallType: b.cfg.NATS.Mode}, ps, registry, b.log)
}
}
+71
View File
@@ -0,0 +1,71 @@
package archied
import (
"fmt"
"maps"
"slices"
"github.com/docker/sandbox-kit-spec/v3/spec"
"github.com/samcharles93/archie-core/internal/config"
"github.com/samcharles93/archie-core/internal/daemon"
"github.com/samcharles93/archie-core/internal/infrastructure/egress"
"github.com/samcharles93/archie-core/internal/storage"
)
const modelProxyCAPath = "/opt/archie/egress-ca.pem"
// modelEgress opens native task containers' model sessions on the egress
// proxy. Each keyed provider's key is the org's credential binding named
// after the provider.
type modelEgress struct {
proxy *egress.Proxy
addr string
caFile string
config *config.Holder
}
func (m modelEgress) Open(token, org string, providers map[string]config.Provider) (daemon.ModelSession, error) {
var keyed []string
for _, name := range slices.Sorted(maps.Keys(providers)) {
if providers[name].APIKeyEnv != "" {
keyed = append(keyed, name)
}
}
if len(keyed) == 0 {
return daemon.ModelSession{}, nil
}
bound := m.config.Get().Containers.BoundCredentials(org, keyed, keyed)
var (
subs []egress.Substitution
hosts []string
env []string
)
for _, name := range keyed {
p := providers[name]
if _, ok := bound[name]; !ok {
return daemon.ModelSession{}, fmt.Errorf("provider %q has no credential binding named %q in org %s", name, name, org)
}
host, err := egress.ProviderHost(p.Class, p.BaseURL)
if err != nil {
return daemon.ModelSession{}, fmt.Errorf("provider %q: %w", name, err)
}
subs = append(subs, egress.Substitution{Service: name, Host: host})
hosts = append(hosts, host)
env = append(env, p.APIKeyEnv+"="+egress.SentinelFor(name))
}
session, err := m.proxy.Register(egress.SessionOptions{
Token: token, Org: org, Substitutions: subs,
Network: &spec.PhasedNetwork{Runtime: &spec.NetworkRules{Allow: hosts}},
})
if err != nil {
return daemon.ModelSession{}, err
}
session.EnterRuntime()
return daemon.ModelSession{
Env: append(env, egress.ModelProxyEnvFor(m.addr, token, hosts, modelProxyCAPath)...),
Mounts: []storage.Mount{{Type: storage.MountTypeBind, Source: m.caFile, Destination: modelProxyCAPath}},
}, nil
}
func (m modelEgress) Close(token string) { m.proxy.Revoke(token) }
+72
View File
@@ -0,0 +1,72 @@
package archied
import (
"context"
"fmt"
"log/slog"
"strings"
"time"
"github.com/samcharles93/archie-core/internal/domain/storecontract"
"github.com/samcharles93/archie-core/internal/domain/workflow/task"
)
const outcomePostInterval = 15 * time.Second
type outcomeDeliverer func(ctx context.Context, channel, chatID, text string) error
// postOutcomes reports each finished chat task to the conversation that
// spawned it. A task stays unposted until delivery succeeds, so a Messaging
// Service that is down or restarting only delays the report.
func postOutcomes(ctx context.Context, tasks storecontract.OutcomePoster, deliver outcomeDeliverer, log *slog.Logger) {
ticker := time.NewTicker(outcomePostInterval)
defer ticker.Stop()
for {
postPendingOutcomes(ctx, tasks, deliver, log)
select {
case <-ctx.Done():
return
case <-ticker.C:
}
}
}
func postPendingOutcomes(ctx context.Context, tasks storecontract.OutcomePoster, deliver outcomeDeliverer, log *slog.Logger) {
pending, err := tasks.UnpostedOutcomes(ctx)
if err != nil {
log.Warn("list unposted task outcomes", "err", err)
return
}
for _, t := range pending {
// An origin that names no chat can never be delivered; marking it
// posted keeps it from being retried forever.
if channel, chatID, ok := splitOrigin(t.Origin); ok {
if err := deliver(ctx, channel, chatID, outcomeText(t)); err != nil {
log.Warn("post task outcome", "task", t.ID, "err", err)
continue
}
}
if err := tasks.MarkOutcomePosted(ctx, t.ID); err != nil {
log.Warn("mark task outcome posted", "task", t.ID, "err", err)
}
}
}
// splitOrigin reads "<channel>:<chat>[/<thread>]", the form messaging.Origin
// writes. Delivery addresses the chat; threads are not addressable.
func splitOrigin(origin string) (channel, chatID string, ok bool) {
channel, conversation, found := strings.Cut(origin, ":")
chatID, _, _ = strings.Cut(conversation, "/")
return channel, chatID, found && channel != "" && chatID != ""
}
func outcomeText(t *task.Task) string {
text := fmt.Sprintf("Task %d (%s) finished: %s", t.ID, t.Title, t.Status)
if t.PRNumber > 0 {
text += fmt.Sprintf(", PR #%d", t.PRNumber)
}
if t.ParkReason != "" {
text += "\n" + t.ParkReason
}
return text
}
+1 -1
View File
@@ -98,5 +98,5 @@ type unscopedBroker struct{}
func (unscopedBroker) Name() string { return "nats_credential_scope" }
func (unscopedBroker) Check(context.Context) health.Result {
return health.Result{Status: health.StatusDegraded, Detail: "external NATS: agent containers receive the broker token, which reaches every subject; use the embedded broker to scope them to their task"}
return health.Result{Status: health.StatusDegraded, Detail: "external NATS: containers get a broker token that reaches every subject; use the embedded broker"}
}
-32
View File
@@ -1,32 +0,0 @@
package archied
import (
"flag"
"fmt"
"io"
"strings"
"github.com/samcharles93/archie-core/internal/releaseupdate"
)
func RunReleaseChannel(args []string, in io.Reader, out, stderr io.Writer) int {
flags := flag.NewFlagSet("select-release", flag.ContinueOnError)
flags.SetOutput(stderr)
channel := flags.String("channel", "stable", "stable, next, or exact-pin")
pin := flags.String("pin", "", "exact release version")
if err := flags.Parse(args); err != nil {
return 1
}
data, err := io.ReadAll(in)
if err != nil {
fmt.Fprintln(stderr, err)
return 1
}
version, err := releaseupdate.SelectRelease(strings.Fields(string(data)), *channel, *pin)
if err != nil {
fmt.Fprintln(stderr, err)
return 1
}
fmt.Fprintln(out, version)
return 0
}
+8 -16
View File
@@ -117,21 +117,8 @@ var reloadableFields = map[string]bool{
"ModelLimits": true,
"Notify": true,
"Review": true,
// BotUser/BotEmail used to be poll-path only; ForTask now carries both
// into the task snapshot as well, because the sandboxed worker builds
// its own worktree.Manager and signs its own commits (agentworker
// task_execution.go, config.go ForTask). The commit identity is
// therefore per-dispatch and does reload. The startup-built daemon
// trees (bootstrap.go:654) still capture both, but only to record
// user.name/user.email in the clone's own .git/config
// (worktree.go setIdentity) -- cosmetic, not the signature go-git
// commits with -- so a reload applies where it counts and this stays
// reloadable rather than requires-restart.
"BotEmail": true,
// cfg.Label additionally feeds IssuesWithLabel per poll cycle
// (daemon.go:359,390) and cfg.BotUser feeds AssignedIssues (:364,377).
"Label": true,
"BotUser": true,
// Poll loops read the dispatch label each cycle.
"Label": true,
// MaxRetries has two consumers, both re-reading it: the webui maxRetriesFor
// handler (api_tasks.go:380,383) reads it per request, and ForTask carries
// it into every dispatched TaskConfig, where the container's remediation
@@ -148,7 +135,6 @@ var reloadableSubFields = map[string]map[string]bool{
"VolumeTTL": true, "Profiles": true, "Credentials": true,
"Image": true, "MaxConcurrency": true, "MaxUptime": true, "PullPolicy": true, "Network": true,
},
"Forge": {"Host": true},
// Policy is carried into TaskConfig by ForTask (config.go); MCPServers,
// WebFetch and Minimax are not and stay requires-restart.
"Tools": {"Policy": true},
@@ -188,3 +174,9 @@ func changedNonReloadableFields(old, next config.Config) []string {
}
return changed
}
// Account credentials, identity namespaces and forge listeners survive a reload.
func preserveForgeIdentity(old, next config.Config) config.Config {
next.BotUser, next.BotEmail, next.Forge = old.BotUser, old.BotEmail, old.Forge
return next
}
@@ -0,0 +1,31 @@
package archied
import (
"slices"
"testing"
"github.com/samcharles93/archie-core/internal/config"
)
func TestReloadPreservesForgeIdentity(t *testing.T) {
for _, tt := range []struct {
name, field string
change func(*config.Config)
}{
{"forge host", "Forge", func(c *config.Config) { c.Forge.Host = "other.example" }},
{"bot user", "BotUser", func(c *config.Config) { c.BotUser = "other-account" }},
{"commit email", "BotEmail", func(c *config.Config) { c.BotEmail = "other@example.test" }},
} {
t.Run(tt.name, func(t *testing.T) {
old := config.Config{BotUser: "archie", BotEmail: "archie@example.test", Forge: config.Forge{Host: "forge.example"}}
next := old.Clone()
tt.change(&next)
if restored := preserveForgeIdentity(old, next); restored.BotUser != old.BotUser || restored.BotEmail != old.BotEmail || restored.Forge != old.Forge {
t.Fatal("reload changed the running forge identity")
}
if changed := changedNonReloadableFields(old, next); !slices.Contains(changed, tt.field) {
t.Fatalf("identity change was reloadable: %v", changed)
}
})
}
}
@@ -2,8 +2,10 @@ package archied
import (
"context"
"slices"
"github.com/samcharles93/archie-core/internal/config"
"github.com/samcharles93/archie-core/internal/daemon"
"github.com/samcharles93/archie-core/internal/domain/workflow"
"github.com/samcharles93/archie-core/internal/infrastructure/egress"
)
@@ -12,30 +14,83 @@ import (
// credential for the run a run credential names, when the request is made,
// from the live grants and bindings. A revoked credential, a grant removed
// mid-run or a binding in another org stops resolving at once.
//
// A setup session's token names no task, so the resolver consults the session
// grants first: the session is bound for the org's bindings of the Kit's
// declared services, which is the intersection the operator's update-on-secret
// authorisation covers.
type runCredentialResolver struct {
runs interface {
TaskForCredential(ctx context.Context, token string) (*workflow.Task, error)
}
sessions interface {
GrantedFor(token string) (org string, services []string, ok bool)
}
// models answers a task's org's own providers. Nil grants only the
// instance providers.
models daemon.TaskModels
config *config.Holder
secrets interface {
Resolve(config.SecretRef) (string, error)
}
// credentials, when set, is told whether each bound secret resolved.
credentials *credentialWatch
}
func (r runCredentialResolver) Resolve(ctx context.Context, credential, service string) (string, error) {
if r.sessions != nil {
if org, services, ok := r.sessions.GrantedFor(credential); ok {
return r.resolveBinding(ctx, org, services, service)
}
}
task, err := r.runs.TaskForCredential(ctx, credential)
if err != nil {
return "", egress.ErrUnbound
}
cfg := r.config.Get()
org, granted := cfg.CredentialAccess(task.Identity)
_, granted := cfg.CredentialAccess(task.Identity)
org := task.OrgOrDefault()
// A configured model provider's key is granted to every run of the org:
// any agent stage needs model access.
if r.isModelProvider(ctx, cfg, task, service) {
granted = append(slices.Clone(granted), service)
}
return r.resolveBinding(ctx, org, granted, service)
}
// resolveBinding resolves service from the org's binding when the run or
// session is granted it. An OAuth binding resolves to the empty string: the
// proxy supplies the org's token set itself.
func (r runCredentialResolver) resolveBinding(ctx context.Context, org string, granted []string, service string) (string, error) {
cfg := r.config.Get()
binding, ok := cfg.Containers.BoundCredentials(org, granted, []string{service})[service]
if !ok {
return "", egress.ErrUnbound
}
if binding.Secret == (config.SecretRef{}) {
// An OAuth binding: the proxy supplies the org's token set itself.
return "", nil
}
return r.secrets.Resolve(binding.Secret)
if r.secrets == nil {
return "", egress.ErrUnbound
}
value, err := r.secrets.Resolve(binding.Secret)
r.credentials.observe(ctx, org, service, err != nil)
return value, err
}
// isModelProvider reports whether service is a provider the task's models
// run on: an instance provider, or one of its org's own.
func (r runCredentialResolver) isModelProvider(ctx context.Context, cfg config.Config, task *workflow.Task, service string) bool {
if _, ok := cfg.Providers[service]; ok {
return true
}
if r.models == nil {
return false
}
models, err := r.models.TaskModels(ctx, task.ID)
if err != nil {
return false
}
_, ok := models.Providers[service]
return ok
}
@@ -29,17 +29,21 @@ func TestEgressResolvesOnlyGrantedCredentials(t *testing.T) {
own := []string{"anthropic"}
cfg := config.Config{
GrantedCredentials: []string{"github"},
Providers: map[string]config.Provider{"openai": {Class: "openai"}, "deepseek": {Class: "deepseek"}},
Identities: []config.IdentityConfig{{Name: "acme-bot", Org: "acme", GrantedCredentials: &own}},
Containers: config.ContainerConfig{Credentials: []config.CredentialBinding{
{Service: "github", Secret: config.SecretRef{Engine: "env", Key: "GH"}},
{Service: "anthropic", Secret: config.SecretRef{Engine: "env", Key: "ROOT_AI"}},
{Service: "anthropic", Org: "acme", Secret: config.SecretRef{Engine: "env", Key: "ACME_AI"}},
{Service: "openai", Secret: config.SecretRef{Engine: "env", Key: "ROOT_OPENAI"}},
{Service: "openai", Org: "acme", Secret: config.SecretRef{Engine: "env", Key: "ACME_OPENAI"}},
{Service: "deepseek", Secret: config.SecretRef{Engine: "env", Key: "ROOT_DEEPSEEK"}},
}},
}
r := runCredentialResolver{
runs: runsByToken{"root-run": {ID: 1}, "acme-run": {ID: 2, Identity: "acme-bot"}},
runs: runsByToken{"root-run": {ID: 1}, "acme-run": {ID: 2, Identity: "acme-bot", Org: "acme"}, "acme-row-run": {ID: 3, Org: "acme"}},
config: config.NewHolder(cfg),
secrets: secretValues{"GH": "gh-secret", "ROOT_AI": "root-ai", "ACME_AI": "acme-ai"},
secrets: secretValues{"GH": "gh-secret", "ROOT_AI": "root-ai", "ACME_AI": "acme-ai", "ROOT_OPENAI": "root-openai", "ACME_OPENAI": "acme-openai", "ROOT_DEEPSEEK": "root-deepseek"},
}
tests := []struct {
name, credential, service, want string
@@ -49,6 +53,10 @@ func TestEgressResolvesOnlyGrantedCredentials(t *testing.T) {
{"identity run gets its own org's binding", "acme-run", "anthropic", "acme-ai"},
{"identity grants replace the root's", "acme-run", "github", ""},
{"unknown credential", "gone", "github", ""},
{"a model provider key needs no grant", "root-run", "openai", "root-openai"},
{"a model provider key is the run's own org's", "acme-run", "openai", "acme-openai"},
{"another org's model provider key is unbound", "acme-run", "deepseek", ""},
{"the task's own org decides, not the identity mapping", "acme-row-run", "openai", "acme-openai"},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
@@ -65,3 +73,47 @@ func TestEgressResolvesOnlyGrantedCredentials(t *testing.T) {
})
}
}
// TestEgressResolvesASessionGrant pins that a setup session's token, which
// names no task, resolves only the granted services of its own org. A grant
// that widened to another org's binding would hand the container a credential
// the operator's update-on-secret authorisation never covered.
func TestEgressResolvesASessionGrant(t *testing.T) {
cfg := config.Config{Containers: config.ContainerConfig{Credentials: []config.CredentialBinding{
{Service: "github", Org: "acme", Secret: config.SecretRef{Engine: "env", Key: "ACME_GH"}},
{Service: "linear", Org: "acme"},
}}}
sessions := &sessionGrants{}
sessions.Grant("acme-session", "acme", []string{"github", "linear"})
sessions.Grant("other-session", "other", []string{"github"})
r := runCredentialResolver{
runs: runsByToken{"acme-run": {ID: 1, Identity: "acme-bot", Org: "acme"}},
sessions: sessions,
config: config.NewHolder(cfg),
secrets: secretValues{"ACME_GH": "acme-gh"},
}
tests := []struct {
name, credential, service, want string
bound bool
}{
{name: "a bound api key resolves", credential: "acme-session", service: "github", want: "acme-gh", bound: true},
{name: "a bound oauth service resolves to the proxy's own token set", credential: "acme-session", service: "linear", bound: true},
{name: "a service outside the grant is unbound", credential: "acme-session", service: "nope"},
{name: "another org's binding is unbound", credential: "other-session", service: "github"},
{name: "an unknown token falls through to the task path", credential: "acme-run", service: "github"},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
got, err := r.Resolve(context.Background(), tt.credential, tt.service)
if !tt.bound {
if !errors.Is(err, egress.ErrUnbound) {
t.Fatalf("got %q, %v; want ErrUnbound", got, err)
}
return
}
if err != nil || got != tt.want {
t.Fatalf("got %q, %v; want %q", got, err, tt.want)
}
})
}
}
+1 -3
View File
@@ -76,7 +76,7 @@ func (b *boot) setupScheduling() error {
}
runners[scheduling.KindWorkflow] = workflowRunner
}
router, err := crondelivery.NewRouter(store, runners, cfg.Events)
router, err := crondelivery.NewRouter(store, runners)
if err != nil {
return fmt.Errorf("scheduling: build router: %w", err)
}
@@ -174,8 +174,6 @@ func (s scheduleResourceStore) MarkRun(ctx context.Context, id string, runAt tim
// The stop path derives its own timeout budget from Background rather than
// the boot context, which is already cancelled by the time shutdown runs --
// the same reasoning as shutdownCuratorRuntime.
//
//nolint:contextcheck
func shutdownSchedulingEngine(e *scheduling.Engine, log *slog.Logger) func() {
return func() {
stopCtx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
+2 -13
View File
@@ -72,18 +72,7 @@ func (b *boot) startWatchdog(ctx context.Context) {
// beginning a fresh poll pass, and therefore how stale the progress marker may
// be before the loop counts as stalled.
//
// It is the root poll interval, or the slowest configured identity in
// multi-identity mode: each identity polls its own repos, identity poll
// intervals are frozen at boot, and the marker is stamped by whichever identity
// polled most recently -- so the worst case between stamps is the slowest
// identity, not the fastest. The root value is re-read per sample, so a
// reloaded poll_interval is honoured.
// The shared interval is re-read so live polling changes are honoured.
func (b *boot) loopCadence() time.Duration {
cadence := b.cfgHolder.Get().PollInterval.Std()
for _, id := range b.d.Identities {
if override := id.Cfg.PollInterval.Std(); override > cadence {
cadence = override
}
}
return cadence
return b.cfgHolder.Get().PollInterval.Std()
}
+1
View File
@@ -17,5 +17,6 @@ func (b *boot) startSecretEngines(ctx context.Context, client *staterpc.Client)
b.applyStatus.Report(ctx, controlplanerpc.ExtensionSettingsKind, version, err)
}
runner := secretengine.Supervise(ctx, b.secrets, source, b.processName, report, b.log)
runner.OnChange(func() { b.publishConfigSnapshot(ctx) })
b.addCleanup(runner.Close)
}
+1 -2
View File
@@ -12,7 +12,6 @@ import (
"github.com/samcharles93/archie-core/internal/app/servicekit"
"github.com/samcharles93/archie-core/internal/domain/presence"
"github.com/samcharles93/archie-core/internal/infrastructure/configuration"
"github.com/samcharles93/archie-core/internal/secret"
)
// RunStatus prints one line per service from the State Store's presence
@@ -31,7 +30,7 @@ func RunStatus(args []string, stdout, stderr io.Writer) int {
fmt.Fprintf(stderr, "archied status: %v\n", err)
return 1
}
client, cleanup, err := servicekit.StateStoreClient(presence.Daemon, doc.Config.Services, secret.NewRegistry())
client, cleanup, err := servicekit.StateStoreClient(presence.Daemon, doc.Config)
if err != nil {
fmt.Fprintf(stderr, "archied status: %v\n", err)
return 1
+2
View File
@@ -48,6 +48,8 @@ func (b *boot) openDaemonWorkflowDefinitions() error {
if err != nil {
return fmt.Errorf("build workflow definitions client: %w", err)
}
definitions.ApplyStatus = b.applyStatus
b.controlPlane.ApplyStatus = b.applyStatus
b.workflowDefinitions = definitions
return nil
}
+199 -43
View File
@@ -7,55 +7,200 @@ import (
"flag"
"fmt"
"io"
"log/slog"
"os"
"os/exec"
"path/filepath"
"regexp"
"strings"
"syscall"
"time"
"github.com/BurntSushi/toml"
"golang.org/x/sys/unix"
"github.com/samcharles93/archie-core/internal/app/servicekit"
"github.com/samcharles93/archie-core/internal/infrastructure/configuration"
"github.com/samcharles93/archie-core/internal/releaseupdate"
)
// RunUpdateTopology observes the host using the candidate's topology contract.
func RunUpdateTopology(args []string, out, stderr io.Writer) int {
flags := flag.NewFlagSet("update-topology", flag.ContinueOnError)
// The watchdog's patience: how long the restarted daemon has to report the
// release, and how long it must then stay healthy.
const (
updateHealthTimeout = 60 * time.Second
updateObserve = 120 * time.Second
updatePoll = 3 * time.Second
)
const updateUsage = `usage:
archied update check [-channel stable|next|exact-pin] [-pin version]
archied update install -version X [-config FILE] [-report FILE -channel C -chat ID -thread ID]
archied update watchdog [-config FILE] [-recover]`
// RunUpdate serves `archied update`: check prints the release snapshot,
// install verifies and stages a release and hands it to the watchdog, and
// watchdog switches over, watches the daemon and commits or rolls back.
func RunUpdate(args []string, out, stderr io.Writer) int {
if len(args) == 0 {
fmt.Fprintln(stderr, updateUsage)
return 2
}
switch args[0] {
case "check":
return runUpdateCheck(args[1:], out, stderr)
case "install":
return runUpdateInstall(args[1:], out, stderr)
case "watchdog":
return runUpdateWatchdog(args[1:], stderr)
default:
fmt.Fprintln(stderr, updateUsage)
return 2
}
}
func runUpdateInstall(args []string, out, stderr io.Writer) int {
flags := flag.NewFlagSet("update install", flag.ContinueOnError)
flags.SetOutput(stderr)
configPath := flags.String("config", "", "bootstrap config file")
releasePath := flags.String("release", "", "release manifest")
configPath := flags.String("config", configuration.DefaultConfigPath(), "bootstrap config file")
version := flags.String("version", "", "release version to install")
releaseURL := flags.String("release-url", releaseupdate.ReleaseURL, "base URL of the release downloads")
var meta releaseupdate.InstallMeta
flags.StringVar(&meta.ReportPath, "report", "", "where the watchdog leaves the outcome")
flags.StringVar(&meta.Channel, "channel", "", "channel the outcome is reported on")
flags.Int64Var(&meta.ChatID, "chat", 0, "chat the outcome is reported to")
flags.IntVar(&meta.ThreadID, "thread", 0, "thread the outcome is reported to")
if err := flags.Parse(args); err != nil {
return 2
}
self, err := os.Executable()
if err != nil {
fmt.Fprintln(stderr, err)
return 1
}
if err := checkUpdateTopology(*configPath, *releasePath, out); err != nil {
ctx := context.Background()
tx, err := hostTransaction(*configPath, filepath.Dir(self), func(line string) { fmt.Fprintln(out, line) })
if err != nil {
fmt.Fprintln(stderr, err)
return 1
}
result, err := tx.Prepare(ctx, releaseupdate.PrepareRequest{
Version: *version, ReleaseURL: *releaseURL, Self: self, Meta: meta,
CheckTopology: func(ctx context.Context, required releaseupdate.Topology, managed bool) error {
return checkUpdateTopology(ctx, *configPath, required, managed, out)
},
})
if err != nil {
fmt.Fprintln(stderr, err)
return 1
}
frozen := filepath.Join(releaseupdate.JournalPath(updateStateDir()), releaseupdate.WatchdogBinary)
if err := spawnWatchdog(ctx, result.RestartRequested, frozen, *configPath, false); err != nil {
// The journal stays: the next archied boot recovers it.
fmt.Fprintln(stderr, err)
return 1
}
if !result.RestartRequested {
fmt.Fprintf(out, "no systemd user manager: the watchdog installs %s without restarting; restart the archie services to run it\n", *version)
}
data, err := json.Marshal(result)
if err != nil {
fmt.Fprintln(stderr, err)
return 1
}
fmt.Fprintln(out, releaseupdate.ResultSentinel+string(data))
return 0
}
func runUpdateWatchdog(args []string, stderr io.Writer) int {
flags := flag.NewFlagSet("update watchdog", flag.ContinueOnError)
flags.SetOutput(stderr)
configPath := flags.String("config", configuration.DefaultConfigPath(), "bootstrap config file")
recovering := flags.Bool("recover", false, "recover an interrupted transaction at boot")
binDir := flags.String("bin-dir", "", "directory of the live binaries")
if err := flags.Parse(args); err != nil {
return 2
}
tx, err := hostTransaction(*configPath, *binDir, func(line string) { fmt.Fprintln(stderr, line) })
if err != nil {
fmt.Fprintln(stderr, err)
return 1
}
if err := tx.Run(context.Background(), *recovering); err != nil {
fmt.Fprintln(stderr, err)
return 1
}
return 0
}
func checkUpdateTopology(configPath, releasePath string, out io.Writer) error {
data, err := os.ReadFile(releasePath)
// hostTransaction wires the transaction to this host: systemd user units,
// Docker, the daemon's health endpoint, the control-plane image policy and
// the State Store's own backup and restore.
func hostTransaction(configPath, binDir string, progress func(string)) (*releaseupdate.Transaction, error) {
_, doc, err := servicekit.Resolve(slog.New(slog.DiscardHandler), configPath, "")
if err != nil {
return nil, err
}
if binDir == "" {
return nil, errors.New("-bin-dir is required")
}
return &releaseupdate.Transaction{
StateDir: updateStateDir(),
BinDir: binDir,
Units: systemdUnits{},
Images: dockerImages{},
Policy: imagePolicy{configPath: configPath},
Database: stateDatabase{configPath: configPath},
Health: daemonHealth{url: doc.Config.Health.URL()},
Progress: progress,
HealthTimeout: updateHealthTimeout,
Observe: updateObserve,
Poll: updatePoll,
}, nil
}
// spawnWatchdog runs the frozen watchdog outside this process, so restarting
// any service, including the one that asked for the update, cannot kill it.
// Under systemd it is a transient unit of its own; without one it is a
// detached session.
func spawnWatchdog(ctx context.Context, managed bool, binary, configPath string, recovering bool) error {
exe, err := os.Executable()
if err != nil {
return err
}
var manifest struct {
Required releaseupdate.Topology `json:"required_topology"`
args := []string{"update", "watchdog", "-config", configPath, "-bin-dir", filepath.Dir(exe)}
if recovering {
args = append(args, "-recover")
}
if err := json.Unmarshal(data, &manifest); err != nil {
stateEnv := "ARCHIE_UPDATE_STATE_DIR=" + updateStateDir()
if managed {
// The units load the env file beside the config for database and
// service credentials; the watchdog needs the same ones.
envFile := "EnvironmentFile=-" + filepath.Join(filepath.Dir(configPath), "env")
run := append([]string{"--user", "--collect", "--quiet", "--unit=archied-update-watchdog", "--setenv=" + stateEnv, "--property=" + envFile, binary}, args...)
if output, err := exec.CommandContext(ctx, "systemd-run", run...).CombinedOutput(); err != nil {
return fmt.Errorf("start the update watchdog: %w: %s", err, output)
}
return nil
}
logFile, err := os.OpenFile(filepath.Join(updateStateDir(), "watchdog.log"), os.O_CREATE|os.O_WRONLY|os.O_APPEND, 0o600)
if err != nil {
return err
}
if len(manifest.Required.Units) == 0 {
return errors.New("release has no required_topology")
defer func() { _ = logFile.Close() }()
cmd := exec.Command(binary, args...) //nolint:noctx // the watchdog must outlive this process
cmd.Env = append(os.Environ(), stateEnv)
cmd.Stdout, cmd.Stderr = logFile, logFile
cmd.SysProcAttr = &syscall.SysProcAttr{Setsid: true}
if err := cmd.Start(); err != nil {
return fmt.Errorf("start the update watchdog: %w", err)
}
for _, unit := range manifest.Required.Units {
if !regexp.MustCompile(`^archie[a-z0-9-]*$`).MatchString(unit) {
return fmt.Errorf("invalid release unit %q", unit)
}
}
data, err = os.ReadFile(configPath)
return cmd.Process.Release()
}
// checkUpdateTopology refuses a host missing a unit or config section the
// release requires. Units are observable only under systemd; without it there
// is nothing to restart, so only the config is checked.
func checkUpdateTopology(ctx context.Context, configPath string, required releaseupdate.Topology, managed bool, out io.Writer) error {
data, err := os.ReadFile(configPath)
if err != nil {
return err
}
@@ -64,25 +209,34 @@ func checkUpdateTopology(configPath, releasePath string, out io.Writer) error {
return fmt.Errorf("observe topology: %w", err)
}
sections := make(map[string]bool)
for _, section := range manifest.Required.ConfigSections {
for _, section := range required.ConfigSections {
sections[section] = hasConfigSection(document, section)
}
ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
defer cancel()
output, observeErr := exec.CommandContext(ctx, "systemctl", "--user", "list-unit-files", "archie*.service", "--no-legend", "--no-pager").Output()
units := observedUnits(output, observeErr)
plan := manifest.Required.MigrationPlan(units, sections)
if len(plan) != 0 {
fmt.Fprintln(out, "Required migration, in order:")
for _, step := range plan {
fmt.Fprintln(out, " -", step)
units := make(map[string]bool)
var observeErr error
if managed {
ctx, cancel := context.WithTimeout(ctx, 10*time.Second)
defer cancel()
var output []byte
output, observeErr = exec.CommandContext(ctx, "systemctl", "--user", "list-unit-files", "archie*.service", "--no-legend", "--no-pager").Output()
units = observedUnits(output, observeErr)
} else {
for _, unit := range required.Units {
units[unit] = true
}
if observeErr != nil {
return fmt.Errorf("topology unknown; apply the printed migration before retrying: %w", observeErr)
}
return errors.New("refusing partial installation; apply the printed migration before retrying")
}
return nil
plan := required.MigrationPlan(units, sections)
if len(plan) == 0 {
return nil
}
fmt.Fprintln(out, "Required migration, in order:")
for _, step := range plan {
fmt.Fprintln(out, " -", step)
}
if observeErr != nil {
return fmt.Errorf("topology unknown; apply the printed migration before retrying: %w", observeErr)
}
return errors.New("refusing partial installation; apply the printed migration before retrying")
}
func hasConfigSection(document map[string]any, section string) bool {
@@ -108,10 +262,11 @@ func updateStateDir() string {
return filepath.Join(dir, "archie", "update")
}
// Recovery runs independently of the daemon unit: it must be able to stop that
// unit and replace its executable without systemd killing the recovery process.
func resumeUpdate(ctx context.Context) (bool, error) {
journal := filepath.Join(updateStateDir(), "transaction")
// resumeUpdate hands an interrupted transaction to the frozen watchdog and
// reports that this boot must stop: recovery may replace this executable and
// restart this unit, so it runs outside it.
func resumeUpdate(ctx context.Context, configPath string) (bool, error) {
journal := releaseupdate.JournalPath(updateStateDir())
if _, err := os.Stat(journal); errors.Is(err, os.ErrNotExist) {
return false, nil
} else if err != nil {
@@ -123,14 +278,15 @@ func resumeUpdate(ctx context.Context) (bool, error) {
}
defer lock.Close()
if err := unix.Flock(int(lock.Fd()), unix.LOCK_EX|unix.LOCK_NB); errors.Is(err, unix.EWOULDBLOCK) {
// A live watchdog owns the transaction and is restarting this unit.
return false, nil
} else if err != nil {
return false, err
}
defer func() { _ = unix.Flock(int(lock.Fd()), unix.LOCK_UN) }()
cmd := exec.CommandContext(ctx, "systemd-run", "--user", "--collect", "--quiet", "--unit=archied-update-recovery", "--setenv=ARCHIE_UPDATE_STATE_DIR="+updateStateDir(), filepath.Join(journal, "watchdog"), "--boot-recover")
if output, err := cmd.CombinedOutput(); err != nil {
return false, fmt.Errorf("schedule update recovery: %w: %s", err, output)
_ = unix.Flock(int(lock.Fd()), unix.LOCK_UN)
frozen := filepath.Join(journal, releaseupdate.WatchdogBinary)
if err := spawnWatchdog(ctx, (systemdUnits{}).Managed(ctx), frozen, configPath, true); err != nil {
return false, fmt.Errorf("schedule update recovery: %w", err)
}
return true, nil
}
+44
View File
@@ -0,0 +1,44 @@
package archied
import (
"context"
"encoding/json"
"flag"
"fmt"
"io"
"os"
"github.com/samcharles93/archie-core/internal/releaseupdate"
)
// runUpdateCheck serves `archied update check`, printing the release snapshot
// as JSON. The channel defaults to the environment.
func runUpdateCheck(args []string, out, stderr io.Writer) int {
flags := flag.NewFlagSet("update check", flag.ContinueOnError)
flags.SetOutput(stderr)
channel := flags.String("channel", envOr("ARCHIE_UPDATE_RELEASE_CHANNEL", "stable"), "stable, next, or exact-pin")
pin := flags.String("pin", os.Getenv("ARCHIE_UPDATE_PIN"), "exact release version")
if err := flags.Parse(args); err != nil {
return 2
}
catalog := &releaseupdate.GitHubCatalog{Channel: func(context.Context) (string, string, error) { return *channel, *pin, nil }}
snapshot, err := catalog.Check(context.Background())
if err != nil {
fmt.Fprintln(stderr, err)
return 1
}
encoder := json.NewEncoder(out)
encoder.SetIndent("", " ")
if err := encoder.Encode(snapshot); err != nil {
fmt.Fprintln(stderr, err)
return 1
}
return 0
}
func envOr(key, fallback string) string {
if value := os.Getenv(key); value != "" {
return value
}
return fallback
}
+101
View File
@@ -0,0 +1,101 @@
package archied
import (
"context"
"encoding/json"
"fmt"
"io"
"net/http"
"os/exec"
"strings"
"time"
"github.com/samcharles93/archie-core/internal/app/statestore"
"github.com/samcharles93/archie-core/internal/domain/health"
)
// systemdUnits drives the archie user units through systemctl --user.
type systemdUnits struct{}
// Managed reports whether a systemd user manager answers.
func (systemdUnits) Managed(ctx context.Context) bool {
return systemctl(ctx, "show-environment") == nil
}
func (systemdUnits) Start(ctx context.Context, unit string) error {
return systemctl(ctx, "restart", unit+".service")
}
func (systemdUnits) Stop(ctx context.Context, unit string) error {
return systemctl(ctx, "stop", unit+".service")
}
func systemctl(ctx context.Context, args ...string) error {
ctx, cancel := context.WithTimeout(ctx, 2*time.Minute)
defer cancel()
output, err := exec.CommandContext(ctx, "systemctl", append([]string{"--user"}, args...)...).CombinedOutput()
if err != nil {
return fmt.Errorf("systemctl --user %s: %w: %s", strings.Join(args, " "), err, strings.TrimSpace(string(output)))
}
return nil
}
// dockerImages pulls through the Docker CLI.
type dockerImages struct{}
func (dockerImages) Pull(ctx context.Context, ref string) error {
ctx, cancel := context.WithTimeout(ctx, 15*time.Minute)
defer cancel()
if output, err := exec.CommandContext(ctx, "docker", "pull", ref).CombinedOutput(); err != nil {
return fmt.Errorf("docker pull %s: %w: %s", ref, err, strings.TrimSpace(string(output)))
}
return nil
}
// stateDatabase is the task database, backed up and restored in-process by
// the State Store's own recovery code.
type stateDatabase struct{ configPath string }
func (d stateDatabase) Backup(ctx context.Context, out string) error {
_, err := statestore.RunRecovery(ctx, statestore.RecoveryOptions{Operation: statestore.RecoveryBackup, Out: out, Config: d.configPath})
return err
}
func (d stateDatabase) Restore(ctx context.Context, from string) error {
_, err := statestore.RunRecovery(ctx, statestore.RecoveryOptions{Operation: statestore.RecoveryRestore, From: from, Config: d.configPath})
return err
}
func (d stateDatabase) SchemaVersion(ctx context.Context) (int64, error) {
return statestore.SchemaVersion(ctx, statestore.RecoveryOptions{Config: d.configPath})
}
// daemonHealth asks archied's /health/detailed, which answers 200 only once
// the daemon serves and every readiness probe passes.
type daemonHealth struct{ url string }
func (h daemonHealth) Healthy(ctx context.Context) (string, error) {
ctx, cancel := context.WithTimeout(ctx, 10*time.Second)
defer cancel()
req, err := http.NewRequestWithContext(ctx, http.MethodGet, h.url+"/health/detailed", nil)
if err != nil {
return "", err
}
resp, err := http.DefaultClient.Do(req)
if err != nil {
return "", err
}
defer func() { _ = resp.Body.Close() }()
body, err := io.ReadAll(io.LimitReader(resp.Body, 1<<20))
if err != nil {
return "", err
}
var report struct {
Status health.Status `json:"status"`
Version string `json:"version"`
}
if resp.StatusCode != http.StatusOK || json.Unmarshal(body, &report) != nil || report.Status != health.StatusOK {
return report.Version, fmt.Errorf("%s: %s", resp.Status, strings.TrimSpace(string(body)))
}
return report.Version, nil
}
+35 -76
View File
@@ -5,54 +5,56 @@ import (
"crypto/sha256"
"encoding/json"
"errors"
"flag"
"fmt"
"io"
"log/slog"
"os"
"path/filepath"
"regexp"
"time"
"github.com/samcharles93/archie-core/internal/app/controlplane"
"github.com/samcharles93/archie-core/internal/app/servicekit"
pb "github.com/samcharles93/archie-core/internal/contracts/controlplane/v1"
"github.com/samcharles93/archie-core/internal/domain/presence"
"github.com/samcharles93/archie-core/internal/secret"
)
type imageChange struct {
Previous string `json:"previous"`
Candidate string `json:"candidate"`
// imagePolicy is the control plane's agent image, reached through the State
// Store contract. Only the image field changes, preserving concurrent edits
// to unrelated settings.
type imagePolicy struct{ configPath string }
func (p imagePolicy) Image(ctx context.Context) (string, error) {
var image string
err := p.with(ctx, func(_ pb.ControlPlaneServiceClient, _ *pb.QueryResponse, _ map[string]json.RawMessage, current string) error {
image = current
return nil
})
return image, err
}
// RunUpdateImage journals and switches the control-plane image policy. It only
// changes the image field, preserving concurrent edits to unrelated settings.
func RunUpdateImage(args []string, stderr io.Writer) int {
flags := flag.NewFlagSet("update-image", flag.ContinueOnError)
flags.SetOutput(stderr)
configPath := flags.String("config", "", "bootstrap config")
journal := flags.String("journal", "", "transaction directory")
candidate := flags.String("prepare", "", "candidate image digest")
rollback := flags.Bool("rollback", false, "restore the prior image")
if err := flags.Parse(args); err != nil {
return 1
}
ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
defer cancel()
if err := updateImage(ctx, *configPath, *journal, *candidate, *rollback); err != nil {
fmt.Fprintln(stderr, err)
return 1
}
return 0
func (p imagePolicy) SetImage(ctx context.Context, expected, desired string) error {
return p.with(ctx, func(rpc pb.ControlPlaneServiceClient, response *pb.QueryResponse, policy map[string]json.RawMessage, current string) error {
if current == desired {
return nil
}
if current != expected {
return errors.New("image policy changed during update; refusing to overwrite it")
}
var err error
if policy["image"], err = json.Marshal(desired); err != nil {
return err
}
data, err := json.Marshal(policy)
if err != nil {
return err
}
_, err = rpc.Command(ctx, &pb.CommandRequest{Kind: controlplane.ContainerRuntimePoliciesKind, Command: "replace", ValueJson: data, ExpectedVersion: response.Resource.Version, Actor: "self-update", Source: "installer", RequestId: fmt.Sprintf("self-update-%x-%d", sha256.Sum256(data), response.Resource.Version)})
return err
})
}
func updateImage(ctx context.Context, configPath, journal, candidate string, rollback bool) error {
_, doc, err := servicekit.Resolve(slog.New(slog.DiscardHandler), configPath, "")
func (p imagePolicy) with(ctx context.Context, use func(pb.ControlPlaneServiceClient, *pb.QueryResponse, map[string]json.RawMessage, string) error) error {
_, doc, err := servicekit.Resolve(slog.New(slog.DiscardHandler), p.configPath, "")
if err != nil {
return err
}
client, cleanup, err := servicekit.StateStoreClient(presence.Daemon, doc.Config.Services, secret.NewRegistry())
client, cleanup, err := servicekit.StateStoreClient(presence.Daemon, doc.Config)
if err != nil {
return err
}
@@ -70,48 +72,5 @@ func updateImage(ctx context.Context, configPath, journal, candidate string, rol
if err := json.Unmarshal(policy["image"], &current); err != nil {
return err
}
path := filepath.Join(journal, "image.json")
if candidate != "" {
return prepareImageChange(path, current, candidate)
}
data, err := os.ReadFile(path)
if err != nil {
return err
}
var change imageChange
if err := json.Unmarshal(data, &change); err != nil {
return err
}
expected, desired := change.Previous, change.Candidate
if rollback {
expected, desired = desired, expected
}
if current == desired {
return nil
}
if current != expected {
return errors.New("image policy changed during update; refusing to overwrite it")
}
policy["image"], err = json.Marshal(desired)
if err != nil {
return err
}
data, err = json.Marshal(policy)
if err != nil {
return err
}
_, err = rpc.Command(ctx, &pb.CommandRequest{Kind: controlplane.ContainerRuntimePoliciesKind, Command: "replace", ValueJson: data, ExpectedVersion: response.Resource.Version, Actor: "self-update", Source: "installer", RequestId: fmt.Sprintf("self-update-%x-%d", sha256.Sum256(data), response.Resource.Version)})
return err
}
func prepareImageChange(path, current, candidate string) error {
if !regexp.MustCompile(`^([a-zA-Z0-9./:_-]+@)?sha256:[a-f0-9]{64}$`).MatchString(candidate) {
return errors.New("candidate image must be pinned by digest")
}
data, err := json.Marshal(imageChange{Previous: current, Candidate: candidate})
if err != nil {
return err
}
return os.WriteFile(path, data, 0o600)
return use(rpc, response, policy, current)
}
+7 -5
View File
@@ -8,18 +8,19 @@ import (
"testing"
"golang.org/x/sys/unix"
"github.com/samcharles93/archie-core/internal/releaseupdate"
)
func TestUpdateTopologyRefusal(t *testing.T) {
root := t.TempDir()
manifest := filepath.Join(root, "release.json")
config := filepath.Join(root, "config.toml")
writeUpdateFixture(t, manifest, `{"required_topology":{"units":["archie-state-store","archie-gateway","archied"],"config_sections":["services.state","services.gateway"]}}`)
required := releaseupdate.Topology{Units: []string{"archie-state-store", "archie-gateway", "archied"}, ConfigSections: []string{"services.state", "services.gateway"}}
writeUpdateFixture(t, config, "[services.state]\ntarget='localhost:9090'\n")
writeUpdateFixture(t, filepath.Join(root, "systemctl"), "#!/bin/sh\nprintf 'archied.service enabled enabled\\n'\n")
t.Setenv("PATH", root+":"+os.Getenv("PATH"))
var output bytes.Buffer
if err := checkUpdateTopology(config, manifest, &output); err == nil {
if err := checkUpdateTopology(t.Context(), config, required, true, &output); err == nil {
t.Fatal("partial topology accepted")
}
for _, missing := range []string{"configure [services.gateway]", "install and enable archie-state-store.service", "install and enable archie-gateway.service"} {
@@ -39,6 +40,7 @@ func TestResumeUpdate(t *testing.T) {
}
called := filepath.Join(root, "called")
writeUpdateFixture(t, filepath.Join(root, "systemd-run"), "#!/bin/sh\nprintf '%s\\n' \"$@\" > '"+called+"'\n")
writeUpdateFixture(t, filepath.Join(root, "systemctl"), "#!/bin/sh\n")
t.Setenv("PATH", root+":"+os.Getenv("PATH"))
lock, err := os.OpenFile(filepath.Join(root, "lock"), os.O_CREATE|os.O_RDWR, 0o600)
if err != nil {
@@ -50,7 +52,7 @@ func TestResumeUpdate(t *testing.T) {
t.Fatal(err)
}
}
resumed, err := resumeUpdate(t.Context())
resumed, err := resumeUpdate(t.Context(), filepath.Join(root, "config.toml"))
if err != nil || resumed == active {
t.Fatalf("resumed=%v err=%v", resumed, err)
}
@@ -59,7 +61,7 @@ func TestResumeUpdate(t *testing.T) {
if err != nil {
t.Fatal(err)
}
if !strings.Contains(string(data), "--boot-recover") {
if !strings.Contains(string(data), "-recover") {
t.Fatalf("not a rollback: %s", data)
}
}
+20 -2
View File
@@ -2,9 +2,14 @@ package archiegateway
import (
"context"
"fmt"
"slices"
"github.com/samcharles93/archie-core/internal/domain/presence"
"github.com/samcharles93/archie-core/internal/infrastructure/gatewayrpc"
"github.com/samcharles93/archie-core/internal/infrastructure/logrpc"
"github.com/samcharles93/archie-core/internal/logging"
"github.com/samcharles93/archie-core/internal/secret"
"github.com/samcharles93/archie-core/internal/skill"
)
@@ -21,8 +26,10 @@ func (b *server) setSkillList(entries []skill.CatalogEntry) {
// channels, so channel reload is left unset and answers Unavailable.
func (b *server) chatCatalog() gatewayrpc.Catalog {
return gatewayrpc.Catalog{
Skills: b.skillEntries,
Curators: b.curatorStates,
CheckSecret: func(ref secret.SecretRef) bool { _, err := b.secrets.Resolve(ref); return err == nil },
Logs: b.recentLogs,
Skills: b.skillEntries,
Curators: b.curatorStates,
}
}
@@ -86,3 +93,14 @@ func skillSource(entry skill.CatalogEntry, workDir, sharedDir string) string {
return "user-global skills"
}
}
func (b *server) recentLogs(ctx context.Context, service string, q logging.Query) (logging.Result, error) {
switch service {
case "", presence.Gateway:
return b.logFeed.Read(q)
case presence.Daemon:
return logrpc.ReadDaemon(ctx, b.taskActionsConn, q)
default:
return logging.Result{}, fmt.Errorf("unknown log service %q", service)
}
}
+3 -6
View File
@@ -7,17 +7,14 @@ import (
// carriesPersonIdentity reports whether a platform's SenderID names a person.
//
// Telegram and email carry a native per-person sender id (a Telegram numeric
// user id, an SMTP from address). The dashboard ("web") has one bearer token,
// not users. Treating a shared token as an identity
// would give it the same standing as a real user, which the read path's
// isolation guarantee depends on never happening.
// Web sender IDs are supplied by the authenticated UI service.
// Telegram and email carry native person identifiers.
//
// An unknown platform fails closed, so a channel added later gets global and
// agent scopes only until someone decides deliberately that it carries people.
func carriesPersonIdentity(platform string) bool {
switch platform {
case "telegram", "email":
case "telegram", "email", "web":
return true
default:
return false
+113 -136
View File
@@ -7,113 +7,46 @@ import (
"strings"
"sync"
"github.com/samcharles93/archie-core/internal/config"
"github.com/samcharles93/archie-core/internal/domain/org"
"github.com/samcharles93/archie-core/internal/gateway"
"github.com/samcharles93/archie-core/internal/infrastructure/modelcatalog"
)
// chatModelManager owns the process-local model selected for interactive chat.
// The configured role-to-model map supplies the available catalog; switching
// chat models does not rewrite daemon configuration or affect workflow roles.
// chatModelManager owns the model selection for interactive chat. Named
// aliases remain defaults; choosing a catalog model does not rewrite them
// or affect workflow steps.
type chatModelManager struct {
mu sync.RWMutex
models []string
active string
// configured, explicit and catalog are the sources the offered set is
// derived from: the role assignments (live via SetConfigured), the
// [chat].models catalog (fixed for the process), and the model catalog's
// own refs (replaced wholesale by SetModelCatalog).
configured map[string]string
explicit []string
catalog []string
mu sync.RWMutex
aliases map[string]string
active map[org.OrgID]string
details map[string]gateway.ModelDetails
providerNames map[string]string
// orgAliases answers the aliases the principal ctx carries may use. Nil
// serves every caller the instance aliases.
orgAliases func(ctx context.Context) (map[string]string, error)
}
// pickChatDefault is the model the manager starts on: the chat role's
// assignment, falling back to the builder role's.
func pickChatDefault(configured map[string]string) string {
if active := strings.TrimSpace(configured["chat"]); active != "" {
return active
}
return strings.TrimSpace(configured["builder"])
}
// mergeModelRefs is the model set the manager offers: the configured role
// assignments plus every catalog, unique and sorted.
func mergeModelRefs(configured map[string]string, catalogs ...[]string) []string {
unique := make(map[string]struct{}, len(configured))
for _, ref := range configured {
ref = strings.TrimSpace(ref)
if ref != "" {
unique[ref] = struct{}{}
}
}
for _, catalog := range catalogs {
for _, ref := range catalog {
ref = strings.TrimSpace(ref)
if ref != "" {
unique[ref] = struct{}{}
}
}
}
models := make([]string, 0, len(unique))
for ref := range unique {
models = append(models, ref)
}
slices.Sort(models)
return models
}
func newChatModelManager(configured map[string]string, catalogs ...[]string) *chatModelManager {
explicit := make([]string, 0, len(catalogs))
for _, catalog := range catalogs {
explicit = append(explicit, catalog...)
}
m := &chatModelManager{
configured: configured,
explicit: explicit,
details: make(map[string]gateway.ModelDetails),
providerNames: make(map[string]string),
}
m.active = pickChatDefault(configured)
m.setModelsLocked()
func newChatModelManager(aliases map[string]string) *chatModelManager {
m := &chatModelManager{active: make(map[org.OrgID]string), details: make(map[string]gateway.ModelDetails), providerNames: make(map[string]string)}
m.SetConfigured(aliases)
return m
}
// setModelsLocked re-derives the offered set from the three sources and keeps
// the active selection when the new set still carries it, falling back the way
// construction does when it does not.
func (m *chatModelManager) setModelsLocked() {
m.models = mergeModelRefs(m.configured, m.explicit, m.catalog)
if !slices.Contains(m.models, m.active) {
m.active = pickChatDefault(m.configured)
}
}
// SetConfigured re-derives the model set the manager offers after a live
// change to model-role-assignments or provider-settings: the list is built
// from the configured roles at construction, so a stored change would leave
// the chat surfaces offering refs the runtime can no longer resolve. The
// operator's active selection survives when the new roles still carry it; a
// removed one falls back the way construction does.
func (m *chatModelManager) SetConfigured(configured map[string]string) {
// SetConfigured replaces the alias table after a live change to model-aliases.
// An available selection survives; otherwise chat returns to the default alias.
func (m *chatModelManager) SetConfigured(aliases map[string]string) {
m.mu.Lock()
defer m.mu.Unlock()
m.configured = configured
m.setModelsLocked()
m.aliases = aliases
m.keepActiveLocked()
}
// SetModelCatalog replaces the model catalog's contribution to the manager:
// the references it offers, the provider display names and the per-model
// details. The previous snapshot's entries are dropped rather than merged --
// a model or provider the catalog no longer publishes must stop being offered
// and described, and these maps are the manager's own state rather than a
// second catalogue to fall back on.
func (m *chatModelManager) SetModelCatalog(snapshot modelcatalog.Snapshot, models []string) {
// SetModelCatalog replaces the provider display names and per-model details
// the catalog publishes.
func (m *chatModelManager) SetModelCatalog(snapshot modelcatalog.Snapshot) {
m.mu.Lock()
defer m.mu.Unlock()
m.catalog = slices.Clone(models)
m.details = make(map[string]gateway.ModelDetails)
m.providerNames = make(map[string]string)
for _, provider := range snapshot.Providers {
@@ -129,7 +62,15 @@ func (m *chatModelManager) SetModelCatalog(snapshot modelcatalog.Snapshot, model
}
}
}
m.setModelsLocked()
m.keepActiveLocked()
}
// keepActiveLocked retains an alias or an available model across live updates.
func (m *chatModelManager) keepActiveLocked() {
selection := m.selectionLocked(org.DefaultOrgID)
if !slices.Contains(config.ChatAliases(m.aliases), selection) && !slices.Contains(chatModelRefs(m.aliases, m.details), selection) {
m.active[org.DefaultOrgID] = config.PurposeChat.DefaultAlias()
}
}
func (m *chatModelManager) ProviderDisplayName(provider string) string {
@@ -145,80 +86,116 @@ func (m *chatModelManager) ModelDetails(ref string) (gateway.ModelDetails, bool)
return details, ok
}
func (m *chatModelManager) Models() []string {
m.mu.RLock()
defer m.mu.RUnlock()
return slices.Clone(m.models)
func (m *chatModelManager) selectionLocked(id org.OrgID) string {
if selection := m.active[id]; selection != "" {
return selection
}
return config.PurposeChat.DefaultAlias()
}
func (m *chatModelManager) ActiveModel() string {
func (m *chatModelManager) ActiveAlias(ctx context.Context) string {
m.mu.RLock()
defer m.mu.RUnlock()
return m.active
selection := m.selectionLocked(org.OrgFromContext(ctx))
if strings.Contains(selection, "/") {
return ""
}
return selection
}
func (m *chatModelManager) SetActiveModel(ctx context.Context, ref string) error {
func (m *chatModelManager) SetActiveModel(ctx context.Context, selection string) error {
if err := ctx.Err(); err != nil {
return err
}
aliases, err := m.aliasesFor(ctx)
if err != nil {
return err
}
m.mu.Lock()
defer m.mu.Unlock()
if !slices.Contains(m.models, ref) {
return fmt.Errorf("model %q is not configured", ref)
id := org.OrgFromContext(ctx)
if id == org.DefaultOrgID || m.orgAliases == nil {
aliases = m.aliases
}
m.active = ref
if !slices.Contains(config.ChatAliases(aliases), selection) && !slices.Contains(m.refsLocked(id, aliases), selection) {
return fmt.Errorf("model or alias %q is not available", selection)
}
m.active[id] = selection
return nil
}
func (m *chatModelManager) Providers() []string {
m.mu.RLock()
defer m.mu.RUnlock()
seen := make(map[string]bool)
var providers []string
for _, model := range m.models {
provider, _, ok := strings.Cut(model, "/")
if ok && provider != "" && !seen[provider] {
seen[provider] = true
providers = append(providers, provider)
}
// aliasesFor is the alias table ctx's org resolves against: the instance's
// for the system org, the org's own answer otherwise.
func (m *chatModelManager) aliasesFor(ctx context.Context) (map[string]string, error) {
if m.orgAliases == nil || org.OrgFromContext(ctx) == org.DefaultOrgID {
m.mu.RLock()
defer m.mu.RUnlock()
return m.aliases, nil
}
slices.Sort(providers)
return providers
return m.orgAliases(ctx)
}
func (m *chatModelManager) ActiveProvider() string {
// ModelsFor returns the model references ctx's org may select.
func (m *chatModelManager) ModelsFor(ctx context.Context) ([]string, error) {
aliases, err := m.aliasesFor(ctx)
if err != nil {
return nil, err
}
m.mu.RLock()
defer m.mu.RUnlock()
provider, _, _ := strings.Cut(m.active, "/")
return provider
id := org.OrgFromContext(ctx)
if id == org.DefaultOrgID || m.orgAliases == nil {
aliases = m.aliases
}
return m.refsLocked(id, aliases), nil
}
func (m *chatModelManager) ModelsForProvider(provider string) []string {
// ModelFor resolves the current selection in the caller's org. A direct
// selection is checked again so revoked models stop serving turns.
func (m *chatModelManager) ModelFor(ctx context.Context) (string, error) {
aliases, err := m.aliasesFor(ctx)
if err != nil {
return "", err
}
m.mu.RLock()
defer m.mu.RUnlock()
prefix := provider + "/"
var models []string
for _, model := range m.models {
if strings.HasPrefix(model, prefix) {
models = append(models, model)
}
id := org.OrgFromContext(ctx)
if id == org.DefaultOrgID || m.orgAliases == nil {
aliases = m.aliases
}
return models
selection := m.selectionLocked(id)
if strings.Contains(selection, "/") {
if !slices.Contains(m.refsLocked(id, aliases), selection) {
return "", fmt.Errorf("model %q is not available in this org", selection)
}
return selection, nil
}
return config.ResolveModel(aliases, selection, config.PurposeChat)
}
func (m *chatModelManager) SetActiveProvider(ctx context.Context, provider string) error {
if err := ctx.Err(); err != nil {
return err
// refsLocked keeps tenant choices within their assigned aliases; the system
// org can also select from its credential-filtered provider catalog.
func (m *chatModelManager) refsLocked(id org.OrgID, aliases map[string]string) []string {
if id != org.DefaultOrgID {
return chatModelRefs(aliases, nil)
}
m.mu.Lock()
defer m.mu.Unlock()
prefix := provider + "/"
for _, model := range m.models {
if strings.HasPrefix(model, prefix) {
m.active = model
return nil
return chatModelRefs(aliases, m.details)
}
func chatModelRefs(aliases map[string]string, catalog map[string]gateway.ModelDetails) []string {
refs := make(map[string]struct{}, len(catalog)+len(aliases))
for ref := range catalog {
refs[ref] = struct{}{}
}
for _, alias := range config.ChatAliases(aliases) {
if ref := strings.TrimSpace(aliases[alias]); ref != "" {
refs[ref] = struct{}{}
}
}
return fmt.Errorf("provider %q is not configured", provider)
out := make([]string, 0, len(refs))
for ref := range refs {
out = append(out, ref)
}
slices.Sort(out)
return out
}
@@ -0,0 +1,119 @@
package archiegateway
import (
"context"
"slices"
"strings"
"testing"
"github.com/samcharles93/archie-core/internal/domain/messaging"
"github.com/samcharles93/archie-core/internal/domain/org"
"github.com/samcharles93/archie-core/internal/gateway"
"github.com/samcharles93/archie-core/internal/infrastructure/modelcatalog"
)
type modelSelectionSessions struct{ gateway.SessionStore }
func (modelSelectionSessions) List(context.Context) ([]gateway.SessionContext, error) {
return nil, nil
}
func TestChatModelSelection(t *testing.T) {
for _, tc := range []struct {
name string
org org.OrgID
selection string
want string
denied bool
}{
{"catalog model", org.DefaultOrgID, "openai/new", "openai/new", false},
{"configured alias", org.DefaultOrgID, "fast", "openai/small", false},
{"unknown model", org.DefaultOrgID, "openai/unknown", "openai/old", true},
{"org model", "tenant", "anthropic/sonnet", "anthropic/sonnet", false},
{"org alias", "tenant", "tenant-chat", "anthropic/sonnet", false},
{"other org model", "tenant", "openai/new", "anthropic/sonnet", true},
{"other org alias", "tenant", "fast", "anthropic/sonnet", true},
} {
t.Run(tc.name, func(t *testing.T) {
manager := newChatModelManager(map[string]string{"default": "openai/old", "fast": "openai/small"})
manager.SetModelCatalog(modelcatalog.Snapshot{Providers: []modelcatalog.Provider{
{ID: "openai", Models: []modelcatalog.Model{{ID: "old"}, {ID: "small"}, {ID: "new"}}},
{ID: "anthropic", Models: []modelcatalog.Model{{ID: "sonnet"}}},
}})
manager.orgAliases = func(context.Context) (map[string]string, error) {
return map[string]string{"default": "anthropic/sonnet", "tenant-chat": "anthropic/sonnet"}, nil
}
ctx := org.WithOrg(t.Context(), tc.org)
router := gateway.NewRouter(nil, nil, "telegram")
router.Models = manager
reply, err := router.Route(ctx, gateway.Inbound{Message: messaging.Message{Text: "/model " + tc.selection}})
if err != nil {
t.Fatal(err)
}
if denied := strings.HasPrefix(reply, "Cannot switch"); denied != tc.denied {
t.Fatalf("reply = %q, denied = %v", reply, tc.denied)
}
got, err := manager.ModelFor(ctx)
if err != nil || got != tc.want {
t.Fatalf("runtime model = %q, %v; want %q", got, err, tc.want)
}
adapter := &gateway.LocalChatAdapter{Router: router, Sessions: modelSelectionSessions{}, Models: manager}
snapshot, err := adapter.Snapshot(ctx)
if err != nil {
t.Fatal(err)
}
provider, _, _ := strings.Cut(tc.want, "/")
if snapshot.ActiveModel != tc.want || snapshot.ActiveProvider != provider || !slices.Contains(snapshot.ModelsByProvider[provider], tc.want) || !slices.Contains(snapshot.Providers, provider) {
t.Fatalf("selector lost model/provider: %+v", snapshot)
}
if tc.org != org.DefaultOrgID && slices.Contains(snapshot.Models, "openai/new") {
t.Fatal("selector exposed another org's model")
}
profile, err := router.Route(ctx, gateway.Inbound{Message: messaging.Message{Text: "/profile"}})
if err != nil || !strings.Contains(profile, tc.want) {
t.Fatalf("profile without task identity = %q, %v", profile, err)
}
})
}
}
func TestChatModelOrgIsolation(t *testing.T) {
for _, selection := range []string{"fast", "tenant/fast"} {
t.Run(selection, func(t *testing.T) {
m := newChatModelManager(map[string]string{"default": "openai/main"})
tenantAliases := map[string]string{"default": "tenant/main", "fast": "tenant/fast"}
m.orgAliases = func(ctx context.Context) (map[string]string, error) {
if org.OrgFromContext(ctx) == "tenant" {
return tenantAliases, nil
}
return map[string]string{"default": "other/main"}, nil
}
tenant := org.WithOrg(t.Context(), "tenant")
if err := m.SetActiveModel(tenant, selection); err != nil {
t.Fatal(err)
}
check := func(systemModel string) {
t.Helper()
for id, want := range map[org.OrgID]string{org.DefaultOrgID: systemModel, "tenant": "tenant/fast", "other": "other/main"} {
got, err := m.ModelFor(org.WithOrg(t.Context(), id))
if err != nil || got != want {
t.Errorf("org %s model = %q, %v; want %q", id, got, err, want)
}
}
}
check("openai/main")
m.SetConfigured(map[string]string{"default": "openai/updated"})
m.SetModelCatalog(modelcatalog.Snapshot{})
check("openai/updated")
delete(tenantAliases, "fast")
if _, err := m.ModelFor(tenant); err == nil {
t.Fatal("revoked tenant choice still executes")
}
adapter := &gateway.LocalChatAdapter{Router: gateway.NewRouter(nil, nil, "telegram"), Sessions: modelSelectionSessions{}, Models: m}
snapshot, err := adapter.Snapshot(tenant)
if err != nil || snapshot.ActiveModel != "" || !slices.Contains(snapshot.ModelsByProvider["tenant"], "tenant/main") {
t.Fatalf("picker cannot recover: %+v, %v", snapshot, err)
}
})
}
}
+7 -7
View File
@@ -57,9 +57,9 @@ func (b *server) setupTranscriber(cfg config.Config, log *slog.Logger) {
})
if ok {
b.transcriber = client
log.Info("voice transcription enabled", "role", transcription.Role)
} else if cfg.Models[transcription.Role] != "" {
log.Warn("voice transcription configured but unavailable; capability disabled", "role", transcription.Role)
log.Info("voice transcription enabled")
} else if _, err := config.ResolveModel(cfg.Models, "", config.PurposeTranscription); err == nil {
log.Warn("voice transcription configured but unavailable; capability disabled")
}
}
@@ -84,18 +84,18 @@ func (b *server) startRateLimiter(ctx context.Context, cfg config.RateLimitConfi
}
// rebuildChatModelRuntime re-derives the gateway chat runtime's provider set
// after a live change to provider-settings or model-role-assignments. The
// after a live change to provider-settings or model-aliases. The
// turn runner reads the runtime through server.chatLLM, so the swap reaches its
// next turn without rebuilding the runner, and the model manager re-derives
// the references it offers from the new role assignments and the catalog.
// the aliases it offers.
func (b *server) rebuildChatModelRuntime(cfg config.Config) {
if b.chatModels == nil {
return // this process serves no chat turns
}
b.setLLM(modelloop.NewRuntime(executionProviders(cfg), cfg.ModelLimits))
b.chatModels.SetConfigured(cfg.Models)
catalog, models := b.catalogState()
b.chatModels.SetModelCatalog(catalog, models)
catalog := b.catalogState()
b.chatModels.SetModelCatalog(catalog)
b.log.Info("chat model runtime rebuilt", "providers", len(cfg.Providers), "models", len(cfg.Models))
}
+35 -33
View File
@@ -15,8 +15,10 @@ import (
"github.com/samcharles93/archie-core/internal/agentexec/modelloop"
"github.com/samcharles93/archie-core/internal/config"
"github.com/samcharles93/archie-core/internal/domain/messaging"
"github.com/samcharles93/archie-core/internal/domain/usage"
"github.com/samcharles93/archie-core/internal/events"
"github.com/samcharles93/archie-core/internal/gateway"
"github.com/samcharles93/archie-core/internal/infrastructure/controlplanerpc"
"github.com/samcharles93/archie-core/internal/installtype"
"github.com/samcharles93/archie-core/internal/releaseupdate"
"github.com/samcharles93/archie-core/internal/tools"
@@ -35,11 +37,13 @@ type chatSetup struct {
// than the value captured when the runner was built.
ToolLimits func() modelloop.ToolLimits
// LLM resolves the provider runtime at each use: a live provider-settings
// or model-role-assignments update swaps the runtime wholesale (ai-sdk's
// or model-aliases update swaps the runtime wholesale (ai-sdk's
// Runtime caches the provider instances it built), so the pointer cannot
// be captured at construction. nil means no model runtime is configured.
LLM func() *runtime.Runtime
ChatModels gateway.ModelManager
LLM func() *runtime.Runtime
ChatModels gateway.ModelManager
// Usage records each model call; nil records nothing.
Usage *usageSink
ToolReg *tools.Registry
Soul gateway.SoulSource
ChatTasks gateway.TaskCreator
@@ -95,23 +99,9 @@ func componentInstallTypeEnricher(nats config.NATSConfig) func(string) (string,
}
}
func makeUpdateService(s chatSetup) *releaseupdate.Service {
cfg := s.Cfg.Get()
if len(cfg.Chat.Telegram.UpdateCheckCommand) == 0 {
return nil
}
updates := &releaseupdate.Service{
Catalog: releaseupdate.CommandCatalog{Command: cfg.Chat.Telegram.UpdateCheckCommand},
StatePath: filepath.Join(cfg.WorkDir, "telegram-update-deferrals.json"),
InstallType: installtype.Type(),
Enrich: componentInstallTypeEnricher(cfg.NATS),
}
if len(cfg.Chat.Telegram.UpdateInstallCommand) != 0 {
updates.Installer = releaseupdate.CommandInstaller{
Command: cfg.Chat.Telegram.UpdateInstallCommand,
HealthURL: cfg.Health.URL(),
}
}
func makeUpdateService(reader controlplanerpc.ResourceReader, cfg config.Config) *releaseupdate.Service {
updates := controlplanerpc.UpdateService(reader, filepath.Join(cfg.WorkDir, "telegram-update-deferrals.json"))
updates.Enrich = componentInstallTypeEnricher(cfg.NATS)
return updates
}
@@ -142,7 +132,7 @@ func newChatTurnRunner(
Sessions: sessionStore,
Models: s.ChatModels,
Soul: s.Soul,
Model: newChatTurnModel(s.LLM, s.ToolReg, cfg.Chat.MaxSteps, s.ToolLimits, s.ProviderOutcomes),
Model: newChatTurnModel(s.LLM, s.ToolReg, cfg.Chat.MaxSteps, s.ToolLimits, s.ProviderOutcomes, s.Usage),
TaskLister: s.ChatTaskLister,
Tasks: s.ChatTasks,
TaskLogs: s.ChatTaskLogs,
@@ -204,31 +194,38 @@ func chatRepoEnv(cfg config.Config, identity string) []gateway.RepoEnv {
// and records at its own adapter (curatorLLMRunner). Anything that reaches the
// runtime without recording leaves /status reporting a stale last-known
// outcome, which is a /status that lies about the provider.
func sendChatTurn(ctx context.Context, llm *runtime.Runtime, chatModel string, options core.GenerateOptions, turn gateway.TurnStream, outcomes *providerOutcomeRecorder, icons map[string]string) (string, error) {
text, err := runChatTurn(ctx, llm, chatModel, options, turn, icons)
func sendChatTurn(ctx context.Context, llm *runtime.Runtime, chatModel string, options core.GenerateOptions, turn gateway.TurnStream, outcomes *providerOutcomeRecorder, usageRecords *usageSink, icons map[string]string) (string, error) {
text, used, err := runChatTurn(ctx, llm, chatModel, options, turn, icons)
usageRecords.record(ctx, usage.SourceChat, chatModel, used)
// record tolerates a nil recorder: a setup built without one (tests, a
// deployment whose health surface is unwired) still makes its calls.
outcomes.record(chatModel, err)
return text, err
}
func runChatTurn(ctx context.Context, llm *runtime.Runtime, chatModel string, options core.GenerateOptions, turn gateway.TurnStream, icons map[string]string) (string, error) {
// runChatTurn returns the reply and the tokens the call used, which a
// failed call may still report.
func runChatTurn(ctx context.Context, llm *runtime.Runtime, chatModel string, options core.GenerateOptions, turn gateway.TurnStream, icons map[string]string) (string, chat.Usage, error) {
if turn == nil {
result, err := llm.Chat(ctx, chatModel, options)
if err != nil {
return "", fmt.Errorf("llm chat: %w", err)
return "", result.TotalUsage, fmt.Errorf("llm chat: %w", err)
}
return result.Text, nil
return result.Text, result.TotalUsage, nil
}
stream, err := llm.ChatStream(ctx, chatModel, options)
if err != nil {
return "", fmt.Errorf("llm chat stream: %w", err)
return "", chat.Usage{}, fmt.Errorf("llm chat stream: %w", err)
}
text := drainChatStream(stream.FullStream, turn, icons)
if _, err := stream.FinishReason(); err != nil {
return "", fmt.Errorf("llm chat stream: %w", err)
var used chat.Usage
if stream.Usage != nil {
used, _ = stream.Usage()
}
return text, nil
if _, err := stream.FinishReason(); err != nil {
return "", used, fmt.Errorf("llm chat stream: %w", err)
}
return text, used, nil
}
// drainChatStream consumes a model stream to close, reporting assistant text
@@ -337,6 +334,7 @@ type chatTitleGenerator struct {
log *slog.Logger
llm func() *runtime.Runtime
chatModels gateway.ModelManager
usage *usageSink
}
// newChatTitleGenerator wires an LLM-backed title generator for a chat
@@ -346,7 +344,7 @@ func newChatTitleGenerator(s chatSetup) gateway.TitleGenerator {
if s.LLM == nil || s.ChatModels == nil {
return nil
}
return &chatTitleGenerator{log: s.Log, llm: s.LLM, chatModels: s.ChatModels}
return &chatTitleGenerator{log: s.Log, llm: s.LLM, chatModels: s.ChatModels, usage: s.Usage}
}
func (g *chatTitleGenerator) GenerateTitle(ctx context.Context, sessionID, firstMessage string) (string, error) {
@@ -369,8 +367,12 @@ func (g *chatTitleGenerator) GenerateTitle(ctx context.Context, sessionID, first
if llm == nil {
return "", fmt.Errorf("no model runtime is configured")
}
text, err := sendChatTurn(ctx, llm, g.chatModels.ActiveModel(),
core.GenerateOptions{Messages: messages, MaxSteps: 1}, nil, nil, nil)
model, err := g.chatModels.ModelFor(ctx)
if err != nil {
return "", err
}
text, err := sendChatTurn(ctx, llm, model,
core.GenerateOptions{Messages: messages, MaxSteps: 1}, nil, nil, g.usage, nil)
if err != nil {
if g.log != nil {
g.log.Error("session title generation failed", "session", sessionID, "err", err)
+5 -14
View File
@@ -4,6 +4,7 @@ import (
"context"
"fmt"
"github.com/samcharles93/archie-core/internal/domain/storecontract"
"github.com/samcharles93/archie-core/internal/domain/taskactions"
"github.com/samcharles93/archie-core/internal/domain/workflow"
"github.com/samcharles93/archie-core/internal/gateway"
@@ -115,7 +116,7 @@ func (a chatTaskListerAdapter) ListChatTasks(ctx context.Context, identity strin
// identity through the tool input.
type chatTaskLogReaderAdapter struct {
tasks func(context.Context, int64) (*workflow.Task, error)
taskLogs *logging.TaskRegistry
taskLogs storecontract.TaskLogStore
}
func (a chatTaskLogReaderAdapter) ReadChatTaskLogs(
@@ -141,19 +142,9 @@ func (a chatTaskLogReaderAdapter) ReadChatTaskLogs(
if attempt <= 0 {
attempt = task.Attempt
}
path := a.taskLogs.Path(taskID, attempt)
if path == "" {
return gateway.ChatTaskLogResult{Attempt: attempt, Entries: []gateway.ChatTaskLogEntry{}}, nil
}
page, err := logging.Page(path, logging.Query{
Component: q.Component,
Contains: q.Contains,
Levels: q.Levels,
Since: q.Since,
Until: q.Until,
Limit: q.Limit,
}, q.AfterID)
page, err := a.taskLogs.TaskLog(ctx, taskID, attempt, logging.Query{
Component: q.Component, Contains: q.Contains, Levels: q.Levels, Since: q.Since, Until: q.Until, Limit: q.Limit, BeforeID: q.AfterID,
})
if err != nil {
return gateway.ChatTaskLogResult{}, err
}
@@ -29,6 +29,7 @@ type chatTurnModel struct {
limits func() modelloop.ToolLimits
// outcomes records each call's result for /status (see sendChatTurn).
outcomes *providerOutcomeRecorder
usage *usageSink
}
func newChatTurnModel(
@@ -37,6 +38,7 @@ func newChatTurnModel(
maxSteps int,
limits func() modelloop.ToolLimits,
outcomes *providerOutcomeRecorder,
usageRecords *usageSink,
) gateway.TurnModel {
return &chatTurnModel{
llm: llm,
@@ -44,6 +46,7 @@ func newChatTurnModel(
maxSteps: maxSteps,
limits: limits,
outcomes: outcomes,
usage: usageRecords,
}
}
@@ -70,6 +73,7 @@ func (m *chatTurnModel) Prepare(
toolInfo: toolSummaries(options.Tools),
toolTokens: gateway.EstimateTokens(string(toolSchema)),
outcomes: m.outcomes,
usage: m.usage,
icons: toolIcons(m.registry),
}, nil
}
@@ -81,6 +85,7 @@ type preparedChatTurnModel struct {
toolInfo []gateway.ToolSummary
toolTokens int
outcomes *providerOutcomeRecorder
usage *usageSink
// icons maps tool name to the icon that tool registered, for chat
// surfaces to render. A nil map yields "" for every lookup, which is
// the no-icon rendering.
@@ -202,5 +207,5 @@ func (m *preparedChatTurnModel) Generate(
if m.llm == nil {
return "", fmt.Errorf("llm chat: no model runtime is configured")
}
return sendChatTurn(ctx, m.llm, m.model, options, stream, m.outcomes, m.icons)
return sendChatTurn(ctx, m.llm, m.model, options, stream, m.outcomes, m.usage, m.icons)
}
+32 -26
View File
@@ -11,7 +11,10 @@ import (
"github.com/samcharles93/ai-sdk/runtime"
"github.com/samcharles93/archie-core/internal/agentexec/modelloop"
"github.com/samcharles93/archie-core/internal/app/controlplane"
"github.com/samcharles93/archie-core/internal/config"
"github.com/samcharles93/archie-core/internal/domain/curator"
"github.com/samcharles93/archie-core/internal/domain/usage"
"github.com/samcharles93/archie-core/internal/events"
infraMemory "github.com/samcharles93/archie-core/internal/infrastructure/memory"
"github.com/samcharles93/archie-core/internal/infrastructure/sessioncurator"
@@ -21,6 +24,11 @@ import (
func (b *server) setupCurators(ctx context.Context) {
log := b.log.With("component", "curator")
model, err := b.chatModels.ModelFor(ctx)
if err != nil {
log.Warn("curators disabled: chat model unavailable", "err", err)
return
}
skillsRoot := b.cfg.SkillsDir
if skillsRoot == "" {
skillsRoot = b.cfg.WorkDir
@@ -32,8 +40,8 @@ func (b *server) setupCurators(ctx context.Context) {
MemoryEngines: b.memEngines,
Skills: skillcurator.NewStore(skillsRoot),
Conversations: sessioncurator.NewAdapter(b.chatSessionStore, b.cfg.BotUser),
LLM: curatorLLMRunner{llm: b.chatLLM, outcomes: b.providerOutcomes},
Model: b.chatModels.ActiveModel(),
LLM: curatorLLMRunner{llm: b.chatLLM, outcomes: b.providerOutcomes, usage: b.usage},
Model: model,
})
if err := b.curatorRegistry.Register(skillcurator.New(skillcurator.DefaultInterval)); err != nil {
@@ -44,30 +52,6 @@ func (b *server) setupCurators(ctx context.Context) {
log.Error("session-memory curator registration failed", "err", err)
}
for _, def := range b.cfg.Curators {
if !def.Enabled {
continue
}
engine := curator.NewDefinitionEngine(curator.Definition{
Name: def.Name,
Enabled: def.Enabled,
Instructions: def.Instructions,
Manifest: curator.Manifest{
Interval: def.Interval.Std(),
Cooldown: def.Cooldown.Std(),
OnInput: def.OnInput,
Tools: def.Tools,
Skills: def.Skills,
MemoryEngine: def.MemoryEngine,
Conversations: def.Conversations,
Model: def.Model,
},
})
if err := b.curatorRegistry.Register(engine); err != nil {
log.Error("config curator registration failed", "curator", def.Name, "err", err)
}
}
b.curatorRuntime = curator.NewRuntime(b.curatorRegistry, curator.RuntimeConfig{})
curator.WakeOnPrimaryInput(ctx, b.bus, b.curatorRuntime, events.KindTurnCompleted)
rt := b.curatorRuntime
@@ -129,6 +113,7 @@ type curatorLLMRunner struct {
// provider is demonstrably reachable -- or worse, keeps showing a much
// older chat outcome as current.
outcomes *providerOutcomeRecorder
usage *usageSink
}
func (r curatorLLMRunner) Chat(ctx context.Context, req curator.ChatRequest) (curator.ChatResult, error) {
@@ -158,6 +143,7 @@ func (r curatorLLMRunner) Chat(ctx context.Context, req curator.ChatRequest) (cu
MaxSteps: max(req.MaxSteps, 1),
})
r.outcomes.record(req.Model, err)
r.usage.record(ctx, usage.SourceCurator, req.Model, res.TotalUsage)
if err != nil {
return curator.ChatResult{}, err
}
@@ -168,3 +154,23 @@ func (r curatorLLMRunner) Chat(ctx context.Context, req curator.ChatRequest) (cu
}
return curator.ChatResult{Text: res.Text, ToolCalls: calls}, nil
}
func (b *server) applyCurators(ctx context.Context, defs []config.CuratorDefinition) error {
definitions := make([]curator.Definition, 0, len(defs))
for _, def := range defs {
definitions = append(definitions, controlplane.CuratorDefinition(def))
}
return b.curatorRuntime.ReplaceDefinitions(ctx, definitions)
}
func (b *server) watchCurators(ctx context.Context, version int64) error {
updates, err := b.controlPlane.WatchCurators(ctx, version)
if err != nil {
return err
}
watchStatus(ctx, b.applyStatus.Report, b.log, controlplane.CuratorsKind, "curators", version, updates, b.controlPlane.WatchCurators,
func(update controlplane.AppliedCurators) int64 { return update.Version },
func(update controlplane.AppliedCurators) error { return update.Err },
func(update controlplane.AppliedCurators) error { return b.applyCurators(ctx, update.Definitions) })
return nil
}
+33 -39
View File
@@ -4,23 +4,27 @@ import (
"context"
"fmt"
"net"
"time"
"google.golang.org/grpc"
"github.com/samcharles93/archie-core/internal/app/controlplane"
"github.com/samcharles93/archie-core/internal/app/servicekit"
"github.com/samcharles93/archie-core/internal/domain/access"
"github.com/samcharles93/archie-core/internal/domain/messaging"
"github.com/samcharles93/archie-core/internal/events"
"github.com/samcharles93/archie-core/internal/gateway"
"github.com/samcharles93/archie-core/internal/infrastructure/gatewayrpc"
"github.com/samcharles93/archie-core/internal/infrastructure/rpcidentity"
"github.com/samcharles93/archie-core/internal/infrastructure/rpctls"
"github.com/samcharles93/archie-core/internal/plugin"
)
// serveGatewayListener is the serving tail of Run: the listener is
// bound, the process declares itself serving, and the chat contract answers
// until ctx ends.
func (b *server) serveGatewayListener(ctx context.Context, listener net.Listener, loopback bool, contract gateway.ChatContract, opts []grpc.ServerOption) error {
func (b *server) serveGatewayListener(ctx context.Context, listener net.Listener, contract gateway.ChatContract, opts []grpc.ServerOption) error {
defer listener.Close()
b.log.Info("archie-gateway running", "addr", listener.Addr().String(), "token_required", !loopback)
b.log.Info("archie-gateway running", "addr", listener.Addr().String(), "tls", b.id != nil)
// Boot is over and this listener is about to accept: the same fact
// systemd's READY=1 asserts, so the announcement goes out here rather than
// from a second notion of "started". Not later: Serve
@@ -66,6 +70,18 @@ func (b *server) startGatewayRuntime(ctx context.Context, actor gateway.ChatTask
if err := b.curatorRuntime.Start(ctx); err != nil {
b.log.Error("curator runtime startup", "err", err)
}
defs, version, err := b.controlPlane.Curators(ctx)
if err == nil {
err = b.applyCurators(ctx, defs)
}
b.applyStatus.Report(ctx, controlplane.CuratorsKind, version, err)
if err != nil {
b.log.Error("curator definitions apply failed", "err", err)
}
if err := b.watchCurators(ctx, version); err != nil {
return nil, fmt.Errorf("watch curators: %w", err)
}
return contract, nil
}
@@ -75,52 +91,30 @@ func (b *server) startGatewayRuntime(ctx context.Context, actor gateway.ChatTask
// rather than rejected as an oversized gRPC message.
const inboundMessageHeadroomBytes = 1 << 20
// gatewayServerOpts applies the transport security boundary: a loopback
// listener is served insecure with no token; any non-loopback address
// requires a Bearer [REDACTED], else the process fails closed. It mirrors
// stateStoreServerOpts, minus the State Store's task-scoped grants: the
// Gateway has one administrative token for all callers.
// gatewayServerOpts serves listen with id's credentials; every caller
// presents a service certificate.
//
// The receive limit is raised above gRPC's 4 MB default because an inbound
// request carries a channel attachment's bytes: the channel frontend
// downloads the file, and this process -- which runs the turn -- holds no
// platform credential to fetch it again.
func gatewayServerOpts(listen, token string) (opts []grpc.ServerOption, loopback bool, err error) {
recv := grpc.MaxRecvMsgSize(messaging.MaxInboundAttachmentBytes + inboundMessageHeadroomBytes)
loopback, err = gatewayrpc.TargetIsLoopback(listen)
func gatewayServerOpts(listen string, id *rpctls.Identity, principals access.PrincipalSource) ([]grpc.ServerOption, error) {
creds, err := id.ServerOption(listen, false)
if err != nil {
return nil, false, err
}
if loopback {
return []grpc.ServerOption{recv}, true, nil
}
if token == "" {
return nil, false, fmt.Errorf(
"gateway listen address %q is non-loopback; non-loopback exposure requires a Bearer [REDACTED] (--token / [services.gateway].target_token) or TLS",
listen,
)
return nil, fmt.Errorf("gateway: %w", err)
}
callers := rpcidentity.Callers{Principals: principals}
return []grpc.ServerOption{
recv,
grpc.ChainUnaryInterceptor(gatewayrpc.UnaryServerInterceptor(token)),
grpc.ChainStreamInterceptor(gatewayrpc.StreamServerInterceptor(token)),
}, false, nil
creds,
grpc.MaxRecvMsgSize(messaging.MaxInboundAttachmentBytes + inboundMessageHeadroomBytes),
grpc.ChainUnaryInterceptor(callers.Unary()),
grpc.ChainStreamInterceptor(callers.Stream()),
}, nil
}
func serveGateway(ctx context.Context, listener net.Listener, contract gateway.ChatContract, sessions gateway.SessionStore, catalog gatewayrpc.Catalog, opts []grpc.ServerOption) error {
server := grpc.NewServer(opts...)
server := grpc.NewServer(append(servicekit.ServerLimits(), opts...)...)
gatewayrpc.RegisterServer(server, contract, sessions, catalog)
serveErr := make(chan error, 1)
go func() { serveErr <- server.Serve(listener) }()
select {
case <-ctx.Done():
// A client can keep a streaming RPC open indefinitely. Allow draining,
// then force cancellation before closing the session store and tools.
timer := time.AfterFunc(10*time.Second, server.Stop)
server.GracefulStop()
timer.Stop()
return nil
case err := <-serveErr:
return err
}
// Serve's bounded drain returns before the session store and tools close.
return servicekit.Serve(ctx, server, listener)
}
+26 -12
View File
@@ -12,8 +12,10 @@ import (
"github.com/samcharles93/archie-core/internal/buildinfo"
"github.com/samcharles93/archie-core/internal/config"
"github.com/samcharles93/archie-core/internal/domain/agent"
"github.com/samcharles93/archie-core/internal/domain/messaging"
"github.com/samcharles93/archie-core/internal/domain/taskactions"
"github.com/samcharles93/archie-core/internal/gateway"
"github.com/samcharles93/archie-core/internal/infrastructure/controlplanerpc"
"github.com/samcharles93/archie-core/internal/taskstate"
"github.com/samcharles93/archie-core/internal/tools"
)
@@ -28,15 +30,15 @@ func (b *server) setupChatRuntime(ctx context.Context, cfg config.Config, actor
// non-command message processing.
providers := executionProviders(cfg)
b.setLLM(modelloop.NewRuntime(providers, cfg.ModelLimits))
// Transcription is a model-role capability, like the chat models it sits
// Transcription is a model-alias capability, like the chat models it sits
// beside: it is built here, on the model-owning side, from this process's
// own [models]/[providers]. The Messaging Service carries a voice note's
// bytes to the Gateway and never holds the credential.
b.setupTranscriber(cfg, b.log)
b.toolReg = tools.NewRegistry()
chatModels := newChatModelManager(cfg.Models, cfg.Chat.Models)
catalog, catalogModels := b.catalogState()
chatModels.SetModelCatalog(catalog, catalogModels)
chatModels := newChatModelManager(cfg.Models)
catalog := b.catalogState()
chatModels.SetModelCatalog(catalog)
b.chatModels = chatModels
// ── Persona registry ─────────────────────────────────────────────
@@ -72,6 +74,14 @@ func (b *server) setupChatRuntime(ctx context.Context, cfg config.Config, actor
// later carries both: the recorder is written by sendChatTurn, the source
// is read by /status.
b.providerOutcomes = newProviderOutcomeRecorder()
b.usage = &usageSink{models: b.chatModels, log: b.log}
if b.stateStore != nil {
b.usage.store = b.stateStore
b.chatModels.orgAliases = func(ctx context.Context) (map[string]string, error) {
_, aliases, err := b.stateStore.CallerModels(ctx)
return aliases, err
}
}
b.statusHealth = newStatusHealth(b)
b.setupChatTasks(cfg)
@@ -86,7 +96,7 @@ func (b *server) setupChatRuntime(ctx context.Context, cfg config.Config, actor
},
cancelExecution: b.stateStore.CancelExecution,
})
b.updateService = makeUpdateService(chatSetup{Cfg: config.NewHolder(cfg)})
b.updateService = makeUpdateService(b.controlPlane, cfg)
return nil
}
@@ -115,7 +125,7 @@ func watchStatus[T any](
open func(ctx context.Context, after int64) (<-chan T, error),
versionOf func(T) int64,
errOf func(T) error,
deliver func(T),
deliver func(T) error,
) {
var lastApplyErr, streamErr error
watchOpen := func(ctx context.Context, after int64) (<-chan T, error) {
@@ -140,7 +150,8 @@ func watchStatus[T any](
log.Error("control-plane watch failed", "kind", kind, "err", err)
return
}
deliver(update)
lastApplyErr = deliver(update)
report(ctx, kind, versionOf(update), lastApplyErr)
},
func() {
// A refusal already on the record outranks the outage; it is the
@@ -171,9 +182,9 @@ func (b *server) watchPersonas(ctx context.Context, version int64) error {
b.controlPlane.WatchPersonas,
func(update controlplane.AppliedPersonas) int64 { return update.Version },
func(update controlplane.AppliedPersonas) error { return update.Err },
func(update controlplane.AppliedPersonas) {
func(update controlplane.AppliedPersonas) error {
b.applyPersonas(update.Collection, update.Version)
b.applyStatus.Report(ctx, controlplane.PersonasKind, update.Version, nil)
return nil
})
return nil
}
@@ -200,9 +211,9 @@ func (b *server) watchSoul(ctx context.Context, version int64) error {
b.controlPlane.WatchSoul,
func(update controlplane.AppliedSoul) int64 { return update.Version },
func(update controlplane.AppliedSoul) error { return update.Err },
func(update controlplane.AppliedSoul) {
func(update controlplane.AppliedSoul) error {
b.applySoul(update.Soul, update.Version)
b.applyStatus.Report(ctx, controlplane.SoulKind, update.Version, nil)
return nil
})
return nil
}
@@ -220,6 +231,7 @@ func (b *server) setupGatewayChat(ctx context.Context, actor gateway.ChatTaskAct
}
cfg := b.cfg
router := gateway.NewRouter(b.stateStore, nil, "web")
router.Settings = messaging.NewSettingsCommand(controlplanerpc.NewSettingsClient(b.stateStore.ControlPlane())).WithIdentities(b.stateStore)
router.Limiter = b.rateLimiter
router.Version = fmt.Sprintf("Archie\nGateway: %s\nRuntime: %s", buildinfo.Version, buildinfo.Runtime)
router.Models = b.chatModels
@@ -237,7 +249,7 @@ func (b *server) setupGatewayChat(ctx context.Context, actor gateway.ChatTaskAct
LLM: b.chatLLM, ChatModels: b.chatModels, ToolReg: b.toolReg,
Soul: b.soul, ChatTasks: b.chatTasks,
ChatTaskLister: chatTaskListerAdapter{tasks: b.stateStore.Tasks},
ChatTaskLogs: chatTaskLogReaderAdapter{tasks: b.stateStore.TaskByID, taskLogs: b.taskLogs},
ChatTaskLogs: chatTaskLogReaderAdapter{tasks: b.stateStore.TaskByID, taskLogs: b.stateStore},
ChatTaskActor: actor,
DefaultChatIdentity: b.defaultChatIdentity,
Bus: b.bus, Log: b.log,
@@ -249,6 +261,7 @@ func (b *server) setupGatewayChat(ctx context.Context, actor gateway.ChatTaskAct
// The Gateway executes the web chat's turns, so it is the process
// that must record their outcomes for /status.
ProviderOutcomes: b.providerOutcomes,
Usage: b.usage,
}
router.LLM, router.LLMStream = makeChatLLMResponder(ctx, "web", setup, b.chatSessionStore, router)
router.Titles = newChatTitleGenerator(setup)
@@ -256,5 +269,6 @@ func (b *server) setupGatewayChat(ctx context.Context, actor gateway.ChatTaskAct
return &gateway.LocalChatAdapter{
Router: router, Sessions: b.chatSessionStore, Turns: gateway.NewTurns(b.log),
Models: b.chatModels, Personas: b.personas, TaskActor: actor, Tasks: b.stateStore,
VoiceAvailable: b.transcriber != nil,
}, nil
}
+2 -5
View File
@@ -4,6 +4,7 @@ import (
"context"
"errors"
"fmt"
"maps"
"reflect"
"slices"
"strings"
@@ -135,11 +136,7 @@ func (b *server) disconnectAbsentMCPServers(ctx context.Context, desired map[str
// applyDesiredMCPServers connects, reconnects or restarts the servers the
// stored document names, skipping the ones already running unchanged.
func (b *server) applyDesiredMCPServers(ctx context.Context, desired map[string]config.MCPServer, state mcpRegistryState) []error {
names := make([]string, 0, len(desired))
for name := range desired {
names = append(names, name)
}
slices.Sort(names)
names := slices.Sorted(maps.Keys(desired))
var problems []error
for _, name := range names {
+5 -3
View File
@@ -8,6 +8,7 @@ import (
"github.com/samcharles93/ai-sdk/core"
"github.com/samcharles93/archie-core/internal/agentexec/modelloop"
"github.com/samcharles93/archie-core/internal/domain/usage"
"github.com/samcharles93/archie-core/internal/tools/mcp"
)
@@ -26,9 +27,9 @@ func (b *server) mcpSamplingHandler() mcp.SamplingHandler {
if b.chatModels == nil {
return mcp.SamplingResult{}, errors.New("mcp sampling: no chat model is configured")
}
model := b.chatModels.ActiveModel()
if model == "" {
return mcp.SamplingResult{}, errors.New("mcp sampling: no chat model is configured")
model, err := b.chatModels.ModelFor(ctx)
if err != nil {
return mcp.SamplingResult{}, fmt.Errorf("mcp sampling: %w", err)
}
messages, err := modelloop.SamplingMessages(req)
if err != nil {
@@ -48,6 +49,7 @@ func (b *server) mcpSamplingHandler() mcp.SamplingHandler {
// recorded for /status like every other one; a nil recorder records
// nothing.
b.providerOutcomes.record(model, err)
b.usage.record(ctx, usage.SourceChat, model, result.TotalUsage)
if err != nil {
return mcp.SamplingResult{}, fmt.Errorf("mcp sampling: %w", err)
}
+42 -24
View File
@@ -21,6 +21,7 @@ import (
"github.com/samcharles93/archie-core/internal/config"
"github.com/samcharles93/archie-core/internal/domain/applystatus"
"github.com/samcharles93/archie-core/internal/domain/curator"
"github.com/samcharles93/archie-core/internal/domain/health"
domainmemory "github.com/samcharles93/archie-core/internal/domain/memory"
"github.com/samcharles93/archie-core/internal/domain/messaging"
"github.com/samcharles93/archie-core/internal/domain/presence"
@@ -28,6 +29,8 @@ import (
"github.com/samcharles93/archie-core/internal/gateway"
"github.com/samcharles93/archie-core/internal/infrastructure/controlplanerpc"
"github.com/samcharles93/archie-core/internal/infrastructure/modelcatalog"
"github.com/samcharles93/archie-core/internal/infrastructure/readiness"
"github.com/samcharles93/archie-core/internal/infrastructure/rpctls"
"github.com/samcharles93/archie-core/internal/infrastructure/secretengine"
"github.com/samcharles93/archie-core/internal/infrastructure/staterpc"
"github.com/samcharles93/archie-core/internal/infrastructure/taskactions"
@@ -49,9 +52,6 @@ type Options struct {
// Listen is the gRPC listen address. Empty falls back to
// [services.gateway].listen.
Listen string
// Token is the bearer token a non-loopback listener requires. Empty falls
// back to [services.gateway].target_token.
Token string
}
// server is the Gateway process: the owner of conversations and of every
@@ -61,9 +61,10 @@ type server struct {
cfgHolder *config.Holder
cfgPath string
log *slog.Logger
taskLogs *logging.TaskRegistry
secrets *secret.Registry
id *rpctls.Identity
engines *secretengine.Runner
stateStore *staterpc.Client
controlPlane *controlplane.Client
applyStatus *applystatus.Reporter
@@ -71,6 +72,7 @@ type server struct {
catalog *modelcatalog.Catalog
chatPool *pgxpool.Pool
chatSessionStore gateway.SessionStore
logFeed *logging.Feed
taskActionsConn *natsio.Conn
bus *events.Bus
@@ -89,6 +91,7 @@ type server struct {
curatorRegistry *curator.Registry
curatorRuntime *curator.Runtime
providerOutcomes *providerOutcomeRecorder
usage *usageSink
statusHealth gateway.HealthSource
updateService *releaseupdate.Service
@@ -126,12 +129,7 @@ func Run(ctx context.Context, options Options) error {
if err != nil {
return err
}
token := options.Token
if token == "" {
token = b.cfg.Services.ResolvedToken(config.ServiceNameGateway, b.secrets.Getenv)
}
//nolint:contextcheck // grpc.StreamServerInterceptor has no context.Context parameter; gatewayrpc.StreamServerInterceptor derives its context from stream.Context() instead
opts, loopback, err := gatewayServerOpts(listen, token)
opts, err := gatewayServerOpts(listen, b.id, b.stateStore)
if err != nil {
return err
}
@@ -139,7 +137,7 @@ func Run(ctx context.Context, options Options) error {
if err != nil {
return fmt.Errorf("listen for gateway: %w", err)
}
return b.serveGatewayListener(ctx, listener, loopback, contract, opts)
return b.serveGatewayListener(ctx, listener, contract, opts)
}
func (b *server) addCleanup(fn func()) { b.cleanups = append(b.cleanups, fn) }
@@ -164,10 +162,11 @@ func (b *server) loadConfig(cfgPath, overlayPath string) error {
b.catalog = modelcatalog.NewCatalog(cfgPath)
b.cfgHolder = config.NewHolder(b.cfg)
logs := servicekit.Logging(b.cfg, "gateway")
b.log, b.taskLogs = logs.Log, logs.TaskLogs
b.log, b.logFeed = logs.Log, logs.Feed
b.addCleanup(func() { _ = logs.Closer.Close() })
b.secrets = secret.NewRegistry()
return nil
b.id, err = servicekit.Identity(b.cfg.TLS, presence.Gateway)
return err
}
// openState claims the conversation store, dials the State Store, starts the
@@ -179,7 +178,7 @@ func (b *server) openState(ctx context.Context) error {
if err := b.claimGatewayOwnership(ctx); err != nil {
return err
}
client, cleanup, err := servicekit.StateStoreClient(presence.Gateway, b.cfg.Services, b.secrets)
client, cleanup, err := servicekit.StateStoreClient(presence.Gateway, b.cfg)
if err != nil {
return fmt.Errorf("state store client: %w", err)
}
@@ -191,7 +190,8 @@ func (b *server) openState(ctx context.Context) error {
report := func(ctx context.Context, version int64, err error) {
b.applyStatus.Report(ctx, controlplanerpc.ExtensionSettingsKind, version, err)
}
b.addCleanup(secretengine.Supervise(ctx, b.secrets, source, applystatus.Gateway, report, b.log).Close)
b.engines = secretengine.Supervise(ctx, b.secrets, source, applystatus.Gateway, report, b.log)
b.addCleanup(b.engines.Close)
cfg, versions, err := b.runtimeConfig(ctx, b.cfg)
if err != nil {
return fmt.Errorf("load runtime settings: %w", err)
@@ -199,12 +199,12 @@ func (b *server) openState(ctx context.Context) error {
b.cfg, b.runtimeVersions = cfg, versions
b.cfgHolder.Set(cfg)
go b.applyStatus.Run(ctx)
go presence.Run(ctx, presence.Gateway, servicekit.Build(), client, nil, b.log)
return nil
}
func (b *server) runtimeConfig(ctx context.Context, base config.Config) (config.Config, map[string]int64, error) {
return servicekit.RuntimeConfig(ctx, b.controlPlane, b.applyStatus, b.secrets, b.log, base, nil)
return servicekit.RuntimeConfig(ctx, b.controlPlane, b.applyStatus, b.secrets, b.log, base, nil, gatewayRuntimeKinds)
}
// connectNATS joins the broker task actions travel over. The daemon owns the
@@ -230,6 +230,16 @@ func (b *server) connectNATS(ctx context.Context) (*natsio.Conn, error) {
}
b.addCleanup(nc.Close)
b.taskActionsConn = nc
registry := health.NewRegistry(
readiness.NewContractProbe("state_store", b.cfg.Health.DependencyTimeout.Std(), func(ctx context.Context) error { _, err := b.stateStore.StatusCounts(ctx); return err }),
readiness.NewContractProbe("nats", b.cfg.Health.DependencyTimeout.Std(), func(context.Context) error {
if !nc.IsConnected() {
return fmt.Errorf("broker disconnected")
}
return nil
}),
)
go presence.Run(ctx, presence.Gateway, servicekit.Build(), b.stateStore, registry, b.log)
return nc, nil
}
@@ -246,7 +256,7 @@ func (b *server) waitForNATSEndpoint(ctx context.Context) (servicekit.NATSEndpoi
}
}
func (b *server) catalogState() (modelcatalog.Snapshot, []string) { return b.catalog.State() }
func (b *server) catalogState() modelcatalog.Snapshot { return b.catalog.State() }
// loadCatalog layers the model catalog under the running config. A catalog
// that cannot be read leaves the configured providers and models in effect.
@@ -256,9 +266,9 @@ func (b *server) loadCatalog(ctx context.Context) {
b.log.Warn("model catalog unavailable; using configured providers and models", "err", err)
return
}
models := modelcatalog.Apply(&b.cfg, snapshot)
modelcatalog.Apply(&b.cfg, snapshot)
b.cfgHolder.Set(b.cfg.Clone())
b.catalog.Set(snapshot, models)
b.catalog.Set(snapshot)
}
func (b *server) startModelCatalogRefresh(ctx context.Context) {
@@ -269,8 +279,8 @@ func (b *server) startModelCatalogRefresh(ctx context.Context) {
return
}
base := b.cfgHolder.Get().Clone()
models := modelcatalog.Apply(&base, snapshot)
b.catalog.Set(snapshot, models)
modelcatalog.Apply(&base, snapshot)
b.catalog.Set(snapshot)
if err := b.relayer(ctx, base, ""); err != nil {
b.log.Warn("model catalog refresh rejected with the settings; the running settings stay", "err", err)
}
@@ -281,11 +291,19 @@ func (b *server) startModelCatalogRefresh(ctx context.Context) {
// chat runtime and tool providers are built from.
var gatewayRuntimeKinds = []string{
controlplane.ProviderSettingsKind,
controlplane.ModelRoleAssignmentsKind,
controlplane.ModelAliasesKind,
controlplane.ToolSettingsKind,
}
func (b *server) startRuntimeWatches(ctx context.Context) error {
// An engine that starts after boot answers references that failed then, so
// the providers resolve again.
b.engines.OnChange(func() {
base := b.cfgHolder.Get()
catalog := b.catalogState()
modelcatalog.Apply(&base, catalog)
_ = b.relayer(ctx, base, controlplane.ProviderSettingsKind)
})
for _, kind := range gatewayRuntimeKinds {
updates, err := b.controlPlane.WatchResource(ctx, kind, b.runtimeVersions[kind])
if err != nil {
@@ -320,7 +338,7 @@ func (b *server) startRuntimeWatches(ctx context.Context) error {
return
}
base := b.cfgHolder.Get()
catalog, _ := b.catalogState()
catalog := b.catalogState()
modelcatalog.Apply(&base, catalog)
lastApplyErr = b.relayer(ctx, base, kind)
},
+15 -7
View File
@@ -40,10 +40,18 @@ func (b *server) refreshSkills(ctx context.Context, applied *string) {
cfg, log := b.cfg, b.log
catalog, err := skill.CatalogRoots(skill.DefaultRoots(cfg.WorkDir, cfg.SkillsDir)...)
if err != nil {
b.applyStatus.Report(ctx, controlplane.SkillsKind, 0, err)
log.Warn("skill catalog load failed", "err", err)
return
}
catalog = append(catalog, b.storedSkills(ctx, catalog)...)
stored, version, err := b.storedSkills(ctx, catalog)
if err != nil {
b.applyStatus.Report(ctx, controlplane.SkillsKind, version, err)
return
}
var applyErr error
defer func() { b.applyStatus.Report(ctx, controlplane.SkillsKind, version, applyErr) }()
catalog = append(catalog, stored...)
b.setSkillList(catalog)
signature := catalogSignature(catalog)
if signature == *applied {
@@ -52,6 +60,7 @@ func (b *server) refreshSkills(ctx context.Context, applied *string) {
b.toolReg.Unregister(skillActivateTool)
if entry := skill.ActivateTool(cfg.WorkDir, catalog); entry != nil {
if err := b.toolReg.Register(*entry); err != nil {
applyErr = err
log.Warn("skill_activate registration failed", "err", err)
return
}
@@ -61,10 +70,10 @@ func (b *server) refreshSkills(ctx context.Context, applied *string) {
}
// storedSkills reads the skills resource, leaving out any name already in
// have. An unreadable resource contributes nothing this round.
func (b *server) storedSkills(ctx context.Context, have []skill.CatalogEntry) []skill.CatalogEntry {
// have. An unreadable resource leaves the running catalog in place.
func (b *server) storedSkills(ctx context.Context, have []skill.CatalogEntry) ([]skill.CatalogEntry, int64, error) {
var out []skill.CatalogEntry
_, _, err := b.controlPlane.Query(ctx, controlplane.SkillsKind, func(value []byte) error {
version, _, err := b.controlPlane.Query(ctx, controlplane.SkillsKind, func(value []byte) error {
var collection controlplane.SkillCollection
if err := json.Unmarshal(value, &collection); err != nil {
return err
@@ -72,8 +81,7 @@ func (b *server) storedSkills(ctx context.Context, have []skill.CatalogEntry) []
for _, held := range collection.Skills {
entry, err := skill.EntryFromContent([]byte(held.Content))
if err != nil {
b.log.Warn("stored skill skipped", "skill", held.Name, "err", err)
continue
return err
}
if !slices.ContainsFunc(have, func(e skill.CatalogEntry) bool { return e.Name == entry.Name }) {
out = append(out, entry)
@@ -84,7 +92,7 @@ func (b *server) storedSkills(ctx context.Context, have []skill.CatalogEntry) []
if err != nil {
b.log.Warn("skills resource unavailable", "err", err)
}
return out
return out, version, err
}
func catalogSignature(catalog []skill.CatalogEntry) string {
+10 -2
View File
@@ -8,11 +8,13 @@ import (
"strings"
"time"
"github.com/samcharles93/ai-sdk/provider/minimax"
"github.com/samcharles93/archie-core/internal/agentexec/modelloop"
"github.com/samcharles93/archie-core/internal/config"
"github.com/samcharles93/archie-core/internal/plugin"
"github.com/samcharles93/archie-core/internal/tools/mcp"
"github.com/samcharles93/archie-core/internal/tools/minimax"
minimaxtool "github.com/samcharles93/archie-core/internal/tools/minimax"
toolprovider "github.com/samcharles93/archie-core/internal/tools/provider"
builtintoolprovider "github.com/samcharles93/archie-core/internal/tools/provider/builtin"
mcptoolprovider "github.com/samcharles93/archie-core/internal/tools/provider/mcp"
@@ -123,7 +125,13 @@ func (b *server) registerMinimaxTool(cfg config.Config, log *slog.Logger) {
return
}
entry := minimax.Tool(minimax.Config{Enabled: true, APIKey: apiKey, BaseURL: cfg.Tools.Minimax.BaseURL})
provider, err := minimax.New(minimax.Config{APIKey: apiKey, BaseURL: cfg.Tools.Minimax.BaseURL})
if err != nil {
log.Warn("minimax video generation enabled but the provider could not be built; tool not registered", "err", err)
return
}
entry := minimaxtool.Tool(provider)
if entry == nil {
return
}
+46
View File
@@ -0,0 +1,46 @@
package archiegateway
import (
"context"
"log/slog"
"time"
"github.com/samcharles93/ai-sdk/chat"
"github.com/samcharles93/archie-core/internal/domain/usage"
"github.com/samcharles93/archie-core/internal/gateway"
)
const usageWriteTimeout = 5 * time.Second
// usageSink records the gateway's model calls. The State Store bills each to
// the org of the principal the call was made for. A nil sink records nothing.
type usageSink struct {
store usage.Recorder
models gateway.ModelManager
log *slog.Logger
}
// record appends one call's usage. It never fails the call: the write is
// detached from the caller's cancellation and a failure is only logged.
func (u *usageSink) record(ctx context.Context, source usage.Source, ref string, used chat.Usage) {
if u == nil || u.store == nil || ref == "" {
return
}
alias := ""
if u.models != nil {
if model, err := u.models.ModelFor(ctx); err == nil && model == ref {
alias = u.models.ActiveAlias(ctx)
}
}
record := usage.Record{
Source: source, Alias: alias,
InputTokens: int64(used.PromptTokens), OutputTokens: int64(used.CompletionTokens),
CachedTokens: int64(used.CachedTokens), At: time.Now().UTC(),
}.WithRef(ref)
ctx, cancel := context.WithTimeout(context.WithoutCancel(ctx), usageWriteTimeout)
defer cancel()
if err := u.store.RecordUsage(ctx, record); err != nil && u.log != nil {
u.log.Warn("record model usage failed", "source", source, "model", ref, "err", err)
}
}
@@ -21,7 +21,7 @@ func channelDescriptors(instances []*channelInstance) []status.Descriptor {
descriptors = append(descriptors, status.Descriptor{
ID: instance.name,
Name: instance.name,
Configured: true,
Configured: instance.current() != nil,
// The capability comes from the same place the action does, so a
// channel cannot declare reload support and then answer nothing.
ReloadSupported: reloadable(instance),
@@ -35,7 +35,14 @@ func channelDescriptors(instances []*channelInstance) []status.Descriptor {
// a reload performs; deriving them separately is how a dashboard ends up offering
// a button that reports success and changes nothing.
func reloadable(instance *channelInstance) bool {
gateway, ok := instance.current().(*telegram.Gateway)
return channelReloadable(instance.current())
}
// channelReloadable reports whether a built channel can re-read its
// configuration in place. A disabled channel (nil) cannot, which is what a
// channel that enablement turned off should declare.
func channelReloadable(ch channels.Channel) bool {
gateway, ok := ch.(*telegram.Gateway)
return ok && gateway.Reload != nil
}
+27 -11
View File
@@ -19,6 +19,7 @@ import (
"github.com/samcharles93/archie-core/internal/domain/messaging"
"github.com/samcharles93/archie-core/internal/domain/storecontract"
"github.com/samcharles93/archie-core/internal/infrastructure/messagingrpc"
"github.com/samcharles93/archie-core/internal/releaseupdate"
"github.com/samcharles93/archie-core/internal/sdnotify"
"github.com/samcharles93/archie-core/internal/secret"
)
@@ -43,6 +44,9 @@ type deps struct {
// SettingsSource enables live channel-settings reconciliation. Nil disables
// it.
SettingsSource chatSettingsSource
// Updates serves /update from the stored update settings. Nil without a
// State Store.
Updates *releaseupdate.Service
// ApplyReporter records what this process applied. Nil reports nothing.
ApplyReporter *applystatus.Reporter
// AppliedVersion is the channel-settings version already running when the
@@ -54,7 +58,7 @@ type deps struct {
}
// channelInstance is one composed channel and what is needed to restart it.
// mu guards channel and cancel.
// mu guards every field below.
type channelInstance struct {
name string
rebuild func(ResolvedConfig) (channels.Channel, error)
@@ -64,6 +68,9 @@ type channelInstance struct {
cancel context.CancelFunc
supervised bool
restartPending bool
// notify wakes a supervisor that is between runs, so an explicit restart
// starts the replacement instead of waiting out a backoff or a park.
notify chan struct{}
}
// current returns the channel instance currently serving, under the lock.
@@ -104,6 +111,9 @@ type Service struct {
appliedVersion atomic.Int64
// reconcileInterval is how often the stored settings are re-read.
reconcileInterval time.Duration
// retry bounds how a supervisor retries a channel that fails on its own
// before parking it for an explicit restart.
retry channelRetryPolicy
mu sync.Mutex
running bool
@@ -131,6 +141,7 @@ func compose(ctx context.Context, d deps) (*Service, error) {
if srv.reconcileInterval <= 0 {
srv.reconcileInterval = applystatus.RestampInterval
}
srv.retry = defaultChannelRetry()
instances, err := composeChannels(ctx, d)
if err != nil {
@@ -149,13 +160,15 @@ func compose(ctx context.Context, d deps) (*Service, error) {
// still restarted alone rather than through a process restart.
func composeChannels(ctx context.Context, d deps) ([]*channelInstance, error) {
instances := make([]*channelInstance, 0, 3)
if d.Config.TelegramToken != "" {
instance, err := composeTelegram(ctx, d)
if err != nil {
return nil, err
}
instances = append(instances, instance)
// Telegram is composed whether or not it is enabled: a channel that
// enablement can turn on has to exist for the supervisor to start. Its
// channel is nil until a token is configured (composeTelegram). The other
// channels are fixed by what the process can build at startup.
telegramInstance, err := composeTelegram(ctx, d)
if err != nil {
return nil, err
}
instances = append(instances, telegramInstance)
for _, instance := range d.ExtensionChannels {
if slices.ContainsFunc(instances, func(have *channelInstance) bool { return have.name == instance.name }) {
d.Log.Warn("channel extension skipped: a built-in channel has the same name", "channel", instance.name)
@@ -163,18 +176,21 @@ func composeChannels(ctx context.Context, d deps) ([]*channelInstance, error) {
}
instances = append(instances, instance)
}
for _, instance := range instances {
instance.notify = make(chan struct{}, 1)
}
return instances, nil
}
func composeTelegram(ctx context.Context, d deps) (*channelInstance, error) {
if len(d.Config.Telegram.AllowedUserIDs) == 0 {
d.Log.Warn("chat.telegram has no allowed_user_ids: every sender will be rejected. " +
"Add your Telegram user id to chat.telegram.allowed_user_ids to enable the bot.")
}
build := func(cfg ResolvedConfig) (channels.Channel, error) {
if cfg.TelegramToken == "" {
return nil, nil
}
if len(cfg.Telegram.AllowedUserIDs) == 0 {
d.Log.Warn("chat.telegram has no allowed_user_ids: every sender will be rejected. " +
"Add your Telegram user id to chat.telegram.allowed_user_ids to enable the bot.")
}
tg := telegram.New(cfg.TelegramToken, cfg.Telegram.AllowedUserIDs, d.Log)
tg.Settings = d.Settings
configureTelegram(ctx, tg, cfg, d)
+16 -37
View File
@@ -8,22 +8,23 @@ import (
"os"
"github.com/samcharles93/archie-core/internal/config"
"github.com/samcharles93/archie-core/internal/domain/presence"
"github.com/samcharles93/archie-core/internal/infrastructure/configuration"
"github.com/samcharles93/archie-core/internal/infrastructure/rpctls"
"github.com/samcharles93/archie-core/internal/secret"
)
// ResolvedConfig contains the resolved inputs needed by the Messaging Service.
type ResolvedConfig struct {
Options Options
Options Options
// Identity is this process's certificate; nil is plaintext, loopback only.
Identity *rpctls.Identity
TelegramToken string
Telegram config.TelegramConfig
// WorkDir, BotUser and HealthURL are not channel transport settings. They
// locate this identity's release-announcement and update-report state
// files, and give the update installer the daemon health endpoint to poll
// after it applies one.
// WorkDir and BotUser are not channel transport settings. They locate
// this identity's release-announcement and update-report state files.
WorkDir string
BotUser string
HealthURL string
ShowToolCalls bool
}
@@ -31,10 +32,10 @@ type projection struct {
gateway ServiceTarget
stateStore ServiceTarget
messaging config.ServiceConnection
tls config.TLS
telegram config.TelegramConfig
workDir string
botUser string
healthURL string
showToolCalls bool
}
@@ -55,17 +56,20 @@ func Resolve(o Options, log *slog.Logger) (ResolvedConfig, error) {
}
}
o = withDefaults(o)
o = withEnvTokens(o)
if err := o.validate(); err != nil {
return ResolvedConfig{}, err
}
id, err := rpctls.Load(proj.tls.CAFile, proj.tls.CertDir, presence.Messaging)
if err != nil {
return ResolvedConfig{}, err
}
return ResolvedConfig{
Options: o,
Identity: id,
Telegram: proj.telegram,
WorkDir: proj.workDir,
BotUser: proj.botUser,
HealthURL: proj.healthURL,
ShowToolCalls: proj.showToolCalls,
}, nil
}
@@ -83,52 +87,27 @@ func readProjection(path, overlay string, log *slog.Logger) (projection, bool, e
func project(cfg config.Config) projection {
return projection{
gateway: ServiceTarget{
Target: cfg.Services.Get(config.ServiceNameGateway).Target,
Token: cfg.Services.Get(config.ServiceNameGateway).TargetToken,
},
stateStore: ServiceTarget{Target: cfg.Services.Get(config.ServiceNameState).Target, Token: cfg.Services.Get(config.ServiceNameState).TargetToken},
gateway: ServiceTarget{Target: cfg.Services.Get(config.ServiceNameGateway).Target},
stateStore: ServiceTarget{Target: cfg.Services.Get(config.ServiceNameState).Target},
messaging: cfg.Services.Get(config.ServiceNameMessaging),
tls: cfg.TLS,
telegram: cfg.Chat.Telegram,
workDir: cfg.WorkDir,
botUser: cfg.BotUser,
healthURL: cfg.Health.URL(),
showToolCalls: cfg.Chat.ShowToolCalls,
}
}
func withEnvTokens(o Options) Options {
if o.Gateway.Token == "" {
o.Gateway.Token = os.Getenv("GATEWAY_TOKEN")
}
if o.StateStore.Token == "" {
o.StateStore.Token = os.Getenv("STATE_STORE_TOKEN")
}
if o.Token == "" {
o.Token = os.Getenv("MESSAGING_TOKEN")
}
return o
}
func merge(o Options, p projection) Options {
if o.Gateway.Target == "" {
o.Gateway.Target = p.gateway.Target
}
if o.Gateway.Token == "" {
o.Gateway.Token = p.gateway.Token
}
if o.StateStore.Target == "" {
o.StateStore.Target = p.stateStore.Target
}
if o.StateStore.Token == "" {
o.StateStore.Token = p.stateStore.Token
}
if o.Listen == "" {
o.Listen = p.messaging.Listen
}
if o.Token == "" {
o.Token = p.messaging.TargetToken
}
return o
}
@@ -0,0 +1,241 @@
package archiemessaging
import (
"context"
"encoding/json"
"log/slog"
"sync"
"testing"
"time"
"github.com/samcharles93/archie-core/internal/channels"
"github.com/samcharles93/archie-core/internal/channels/status"
"github.com/samcharles93/archie-core/internal/config"
"github.com/samcharles93/archie-core/internal/domain/messaging"
"github.com/samcharles93/archie-core/internal/secret"
)
// servingChannel reports itself running once it starts, which is what an
// enable is reported active on: a channel that only reaches "starting" is not
// serving yet.
type servingChannel struct {
mu sync.Mutex
starts int
}
func (c *servingChannel) Name() string { return "telegram" }
func (c *servingChannel) Start(ctx context.Context, _ messaging.ChatContract, lifecycle channels.Lifecycle) error {
c.mu.Lock()
c.starts++
c.mu.Unlock()
if lifecycle.Starting != nil {
lifecycle.Starting()
}
if lifecycle.Running != nil {
lifecycle.Running()
}
<-ctx.Done()
return nil
}
func (c *servingChannel) Stop(context.Context) error { return nil }
func (c *servingChannel) ConfigSchema() json.RawMessage { return nil }
func (c *servingChannel) ValidateConfig(map[string]any) error { return nil }
func (c *servingChannel) count() int {
c.mu.Lock()
defer c.mu.Unlock()
return c.starts
}
// waitCount blocks until count reaches want, so an async start is observed
// rather than slept for.
func waitCount(t *testing.T, count func() int, want int) {
t.Helper()
deadline := time.Now().Add(5 * time.Second)
for time.Now().Before(deadline) {
if count() >= want {
return
}
time.Sleep(time.Millisecond)
}
t.Fatalf("channel started %d times, want at least %d", count(), want)
}
// startEnablementService composes and starts a service with the Telegram
// channel's factory replaced by a scripted channel, so enablement can be driven
// without a real bot.
func startEnablementService(t *testing.T, ch *servingChannel) *Service {
t.Helper()
srv, err := compose(context.Background(), deps{Log: slog.Default()})
if err != nil {
t.Fatalf("compose: %v", err)
}
inst := srv.instanceByID("telegram")
if inst == nil {
t.Fatal("telegram channel was not composed")
}
inst.rebuild = func(ResolvedConfig) (channels.Channel, error) { return ch, nil }
srv.retry = channelRetryPolicy{attempts: 3, backoff: time.Millisecond, maxBackoff: time.Millisecond}
ctx, cancel := context.WithCancel(context.Background())
done := make(chan struct{})
go func() {
defer close(done)
_ = srv.Start(ctx)
}()
t.Cleanup(func() {
cancel()
select {
case <-done:
case <-time.After(5 * time.Second):
t.Error("service did not stop after cancel")
}
})
waitSupervised(t, srv, "telegram")
return srv
}
// waitSupervised blocks until the channel's supervisor exists, so an enable
// issued right after Start is not refused for racing it.
func waitSupervised(t *testing.T, srv *Service, id string) {
t.Helper()
deadline := time.Now().Add(5 * time.Second)
for time.Now().Before(deadline) {
inst := srv.instanceByID(id)
inst.mu.Lock()
supervised := inst.supervised
inst.mu.Unlock()
if supervised {
return
}
time.Sleep(time.Millisecond)
}
t.Fatalf("channel %q was never supervised", id)
}
func channelByID(t *testing.T, srv *Service, id string) status.Status {
t.Helper()
for _, st := range srv.ChannelStatus() {
if st.ID == id {
return st
}
}
t.Fatalf("no channel %q in status", id)
return status.Status{}
}
// TestChannelEnablementAppliesLive pins that enabling and disabling a channel
// starts and stops it without a service restart, and that its declaration
// follows: a channel that was never configured is not reported as configured,
// and one that enablement stopped is reported stopped and unconfigured.
func TestChannelEnablementAppliesLive(t *testing.T) {
ch := &servingChannel{}
srv := startEnablementService(t, ch)
if before := channelByID(t, srv, "telegram"); before.Configured || before.State != status.StateStopped {
t.Fatalf("telegram before enable = %+v, want not configured and stopped", before)
}
if err := srv.enableChannel("telegram", ResolvedConfig{TelegramToken: "token"}); err != nil {
t.Fatalf("enable: %v", err)
}
waitCount(t, ch.count, 1)
if enabled := channelByID(t, srv, "telegram"); !enabled.Configured || enabled.State != status.StateRunning {
t.Fatalf("telegram after enable = %+v, want configured and running", enabled)
}
if err := srv.disableChannel("telegram"); err != nil {
t.Fatalf("disable: %v", err)
}
if disabled := channelByID(t, srv, "telegram"); disabled.Configured || disabled.State != status.StateStopped {
t.Fatalf("telegram after disable = %+v, want not configured and stopped", disabled)
}
}
// fakeSettings serves a stored chat config at a version, the way the
// control-plane client does.
type fakeSettings struct {
chat config.ChatConfig
version int64
}
func (f fakeSettings) RuntimeChatConfig(context.Context, config.ChatConfig) (config.ChatConfig, int64, error) {
return f.chat, f.version, nil
}
// fakeEngine resolves every key, so a stored token reference is resolvable.
type fakeEngine struct{}
func (fakeEngine) Name() string { return "test" }
func (fakeEngine) Resolve(key string) (string, error) { return "token-" + key, nil }
// TestReconcileOnceEnablesAndDisables drives the whole stored-settings path:
// a settings version that configures a token enables the channel and is
// recorded as applied only once the channel is running it; a later version
// that clears the token disables it.
func TestReconcileOnceEnablesAndDisables(t *testing.T) {
ch := &servingChannel{}
srv := startEnablementService(t, ch)
srv.secrets = secret.NewRegistry()
srv.secrets.Register(fakeEngine{})
srv.settingsSource = fakeSettings{
chat: config.ChatConfig{Telegram: config.TelegramConfig{Token: secret.SecretRef{Engine: "test", Key: "bot"}}},
version: 1,
}
if err := srv.reconcileOnce(context.Background()); err != nil {
t.Fatalf("reconcile enable: %v", err)
}
waitCount(t, ch.count, 1)
if got := channelByID(t, srv, "telegram"); !got.Configured {
t.Fatalf("telegram not enabled by reconcile: %+v", got)
}
if srv.appliedVersion.Load() != 1 {
t.Fatalf("appliedVersion = %d after enable, want 1", srv.appliedVersion.Load())
}
srv.settingsSource = fakeSettings{version: 2}
if err := srv.reconcileOnce(context.Background()); err != nil {
t.Fatalf("reconcile disable: %v", err)
}
if got := channelByID(t, srv, "telegram"); got.Configured || got.State != status.StateStopped {
t.Fatalf("telegram not disabled by reconcile: %+v", got)
}
if srv.appliedVersion.Load() != 2 {
t.Fatalf("appliedVersion = %d after disable, want 2", srv.appliedVersion.Load())
}
}
// TestApplyChannelSettingsEnablesAndDisables pins the reconcile entry point:
// a settings change that turns a channel on or off starts or stops it and
// reports applied, and a tick that changes nothing is a no-op.
func TestApplyChannelSettingsEnablesAndDisables(t *testing.T) {
ch := &servingChannel{}
srv := startEnablementService(t, ch)
applied, err := srv.applyChannelSettings(ResolvedConfig{TelegramToken: "token"})
if err != nil || !applied {
t.Fatalf("enable: applied=%v err=%v", applied, err)
}
waitCount(t, ch.count, 1)
if got := channelByID(t, srv, "telegram"); !got.Configured || got.State == status.StateStopped {
t.Fatalf("telegram not enabled: %+v", got)
}
applied, err = srv.applyChannelSettings(ResolvedConfig{})
if err != nil || !applied {
t.Fatalf("disable: applied=%v err=%v", applied, err)
}
if got := channelByID(t, srv, "telegram"); got.Configured || got.State != status.StateStopped {
t.Fatalf("telegram not disabled: %+v", got)
}
applied, err = srv.applyChannelSettings(ResolvedConfig{})
if err != nil || !applied {
t.Fatalf("no-op tick: applied=%v err=%v", applied, err)
}
}
+2 -39
View File
@@ -4,9 +4,6 @@ package archiemessaging
import (
"errors"
"fmt"
"net"
"strings"
"time"
)
@@ -18,11 +15,9 @@ const (
defaultShutdownTimeout = 5 * time.Second
)
// ServiceTarget is one remote contract's endpoint and the service-to-service
// bearer token this process presents to it.
// ServiceTarget is one remote contract's endpoint.
type ServiceTarget struct {
Target string
Token string
}
// Options contains the process inputs for the Messaging Service.
@@ -32,10 +27,8 @@ type Options struct {
Gateway ServiceTarget
StateStore ServiceTarget
// Listen is where MessagingService is served; Token is the bearer token
// its callers present on a non-loopback address.
// Listen is where MessagingService is served.
Listen string
Token string
DependencyTimeout time.Duration
ShutdownTimeout time.Duration
}
@@ -60,35 +53,5 @@ func (o Options) validate() error {
if o.Gateway.Target == "" {
return errors.New("gateway target is required (pass -gateway-target or configure [services.gateway].target)")
}
if err := validateServiceTarget("gateway", o.Gateway); err != nil {
return err
}
if o.StateStore.Target != "" {
if err := validateServiceTarget("state store", o.StateStore); err != nil {
return err
}
}
return validateServiceTarget("messaging listen", ServiceTarget{Target: o.Listen, Token: o.Token})
}
func validateServiceTarget(serviceName string, target ServiceTarget) error {
host, _, err := net.SplitHostPort(target.Target)
if err != nil {
host = target.Target
}
if isLoopback(host) {
return nil
}
if target.Token == "" {
return fmt.Errorf("%s target %q is non-loopback and requires a bearer token", serviceName, target.Target)
}
return nil
}
func isLoopback(host string) bool {
if host == "localhost" {
return true
}
ip := net.ParseIP(strings.Trim(host, "[]"))
return ip != nil && ip.IsLoopback()
}
+36 -30
View File
@@ -85,30 +85,32 @@ func resolveChatSecrets(base ResolvedConfig, layered config.ChatConfig, secrets
return next, nil
}
// applyChannelSettings restarts only channels whose transport settings
// changed and reports whether every restart succeeded. Enabling or
// disabling a channel or changing a listen address needs a process restart.
// applyChannelSettings applies the changed channel settings: channels whose
// enablement changed are started or stopped, channels whose transport settings
// changed are restarted, and the report is success only once every change is in
// effect.
func (s *Service) applyChannelSettings(next ResolvedConfig) (bool, error) {
current := s.currentConfig()
effective, outstanding := pinProcessBindings(current, next)
if next.ShowToolCalls != current.ShowToolCalls {
s.applyShowToolCalls(next.ShowToolCalls)
}
applied := true
for _, id := range channelChanges(current, effective) {
if err := s.restartChannel(id, effective); err != nil {
outstanding := s.applyChannelEnablement(current, next)
if len(outstanding) > 0 {
applied = false
}
for _, id := range channelChanges(current, next) {
if err := s.restartChannel(id, next); err != nil {
outstanding = append(outstanding, err.Error())
applied = false
}
}
// A failed restart leaves the running configuration in place, so the next
// tick sees the same change and retries it. A change that was carried out
// (including a binding change that was refused and reported) is recorded, so
// it is not applied again on every tick.
// A failed change leaves the running configuration in place, so the next
// tick sees the same difference and retries it.
if applied {
s.setConfig(effective)
s.setConfig(next)
}
if len(outstanding) == 0 {
return applied, nil
@@ -116,6 +118,24 @@ func (s *Service) applyChannelSettings(next ResolvedConfig) (bool, error) {
return applied, errors.New(strings.Join(outstanding, "; "))
}
// applyChannelEnablement starts or stops the channels whose enablement the
// stored settings changed, and reports only once the requested state is in
// effect. A channel whose enablement did not change is left alone.
func (s *Service) applyChannelEnablement(current, next ResolvedConfig) []string {
var outstanding []string
switch {
case current.TelegramToken != "" && next.TelegramToken == "":
if err := s.disableChannel("telegram"); err != nil {
outstanding = append(outstanding, err.Error())
}
case current.TelegramToken == "" && next.TelegramToken != "":
if err := s.enableChannel("telegram", next); err != nil {
outstanding = append(outstanding, err.Error())
}
}
return outstanding
}
// applyShowToolCalls updates the running Telegram gateways in place. Nothing
// else consumes the setting, and it is read per reply, so it needs no restart.
func (s *Service) applyShowToolCalls(show bool) {
@@ -126,28 +146,14 @@ func (s *Service) applyShowToolCalls(show bool) {
}
}
// pinProcessBindings holds the values a running process cannot change in place:
// the set of channels it composed.
// The returned configuration keeps the running value for each, and the returned
// problems name what the operator changed and when it takes effect.
func pinProcessBindings(current, next ResolvedConfig) (ResolvedConfig, []string) {
effective := next
var problems []string
if (current.TelegramToken != "") != (next.TelegramToken != "") {
problems = append(problems, "chat.telegram enabled/disabled; takes effect on restart")
effective.Telegram, effective.TelegramToken = current.Telegram, current.TelegramToken
}
return effective, problems
}
// channelChanges returns the composed channels whose transport settings differ
// between the running and the effective configuration. A channel absent from
// current (not composed) never appears, and an unchanged channel is not
// restarted.
// between the running and the next configuration while remaining enabled. A
// channel absent from current (not composed) never appears, an unchanged
// channel is not restarted, and an enable or disable is applied as enablement
// rather than as a transport change.
func channelChanges(current, next ResolvedConfig) []string {
var changed []string
if current.TelegramToken != "" &&
if current.TelegramToken != "" && next.TelegramToken != "" &&
(current.TelegramToken != next.TelegramToken || !slices.Equal(current.Telegram.AllowedUserIDs, next.Telegram.AllowedUserIDs)) {
changed = append(changed, "telegram")
}
Loaded 100 of 890 files, more files were not shown because too many files have changed in this diff. Show more