Commit Graph
101 Commits
Author SHA1 Message Date
sam 892ed291b9 refactor(mcp): use the official MCP Go SDK for the client
Deletes the hand-written internal/tools/mcp client and transports. The
provider talks to an SDK ClientSession over CommandTransport,
StreamableClientTransport or SSEClientTransport, and reconnects when the
server exits. The http transport now keeps a real streamable session
instead of stateless POSTs. Sampling goes through one adapter, since the
SDK marks it deprecated (SEP-2577). message_endpoint is removed: the
SSE transport takes the POST URL from the server's endpoint event.
2026-11-13 03:44:55 +11:00
sam db346b159d fix(webui): mark cookies Secure behind a TLS proxy; run gosec in the gate (#1919)
Cookies took Secure from r.TLS, which is nil behind a proxy that terminates
TLS, so sign-in cookies went out without it. Every dashboard cookie now goes
through one builder that reads the scheme the way the origin check does.

The worktree reset and the agent worker's chown walk use os.Root, so a
symlink swapped in mid-walk can't point them outside the worktree.

gosec runs in golangci-lint, tests included. G104, G115, G204 and G304 are
excluded in config; each remaining finding is fixed or suppressed at the site
with its reason.
2026-11-13 03:16:15 +11:00
sam f92c5b4536 chore(runtime): delete Kit and external agent harness support (#1877)
ci/woodpecker/pr/quality Pipeline was successful
Quality / check (pull_request) Canceled after 7m39s
Archie runs its own agent loop only. Removes Kit resolution and launch,
the harness runner and adapters, the setup terminal and sandbox view,
harness secrets and OAuth capture (table dropped in migration 0075), the
task-tools contract that only Kit containers reached, and the
sandbox-kit-spec dependency. Agent profiles carry image and tools only.
The egress proxy keeps the model-session path; the sandbox relay code
stays for #1867.
2026-11-12 21:29:05 +11:00
sam 44237b448b refactor(prompts): every model prompt is a file in internal/prompts
deploy / archied-image (push) Canceled after 0s
deploy / agent-image (push) Canceled after 0s
deploy / dist-zip (push) Canceled after 0s
deploy / plan (push) Successful in 6s
ci/woodpecker/push/deploy Pipeline was canceled
deploy / gate (push) Canceled after 2m58s
2026-11-11 00:01:43 +11:00
sam 5ea115746c fix(dashboard): blocked tasks say what they need, and a provider refusal names its fix
deploy / plan (push) Successful in 6s
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 5m37s
ci/woodpecker/push/deploy Pipeline was canceled
2026-11-10 23:19:29 +11:00
sam dc2c3b983f feat(workflow): a budget continuation resumes at the stopped step with its conversation (#1832)
deploy / plan (push) Successful in 7s
deploy / gate (push) Successful in 8m12s
deploy / agent-image (push) Successful in 28s
deploy / dist-zip (push) Skipped
deploy / archied-image (push) Successful in 2m11s
2026-11-09 12:44:01 +11:00
sam 45b368af2e docs(agentexec): drop history and refactoring narration from comments
deploy / plan (push) Failing after 2s
deploy / gate (push) Skipped
deploy / archied-image (push) Skipped
deploy / agent-image (push) Skipped
deploy / dist-zip (push) Skipped
2026-11-09 01:45:07 +11:00
sam 0f6448c048 refactor(agentexec): delete unused runner injection seams, ServeCaptureMCP, an unreachable approval branch and orphan fixtures
deploy / plan (push) Failing after 1s
deploy / gate (push) Skipped
deploy / archied-image (push) Skipped
deploy / agent-image (push) Skipped
deploy / dist-zip (push) Skipped
2026-11-09 01:43:53 +11:00
sam b15858adc9 fix(agentexec): clip tool detail on a rune boundary, stop resuming without a session, keep captures when the ready marker fails, log with context
deploy / plan (push) Failing after 1s
deploy / gate (push) Skipped
deploy / archied-image (push) Skipped
deploy / agent-image (push) Skipped
deploy / dist-zip (push) Skipped
2026-11-09 01:41:46 +11:00
sam 1ef576ca7b fix(agentexec): run gate commands as the harness user and reap the process group after each invocation
deploy / plan (push) Failing after 7s
deploy / gate (push) Skipped
deploy / archied-image (push) Skipped
deploy / agent-image (push) Skipped
deploy / dist-zip (push) Skipped
2026-11-09 01:40:10 +11:00
sam 69a46249cc feat(webui): draw each transcript entry with its Trace pattern (#1718)
Quality / check (pull_request) Successful in 3m38s
Record the model's reasoning parts. The step run view shows reasoning as a
Thought row, shell calls as a terminal, the gate as a test run, web_fetch as a
source, finish as the verdict, and each fan-out branch as a subagent.
2026-11-01 22:09:53 +11:00
sam 6f10ecc04f feat(agentexec): record each agent run's transcript and show it per step (#1718)
The loop runner hooks every model call and tool call and reports ordered,
redacted, clipped entries. Each is persisted as an agent_transcript task
event through InsertEvent, left out of the timeline read, and served per
attempt at /api/tasks/{id}/transcript. The step run view shows it with Trace.
2026-11-01 22:09:53 +11:00
sam 0611e00330 feat(llm): trace model and tool calls via ai-sdk v0.5.0 2026-10-25 23:19:40 +11:00
sam f2b8681611 refactor(workflow): stop assuming a human decides
An approval, ask or park can be resolved by an operator, a rule, a
classifier or automatically, so the vocabulary stops naming a person:

- step type human.approve -> workflow.approve
- park class needs_human -> needs_decision (ParkNeedsDecision)
- event kinds human_* -> operator_*
- alert kind task.needs_human -> task.needs_decision
- status grouping "needs you" -> "needs attention"

0061 rewrites the stored values, including the approval step type in the
workflow-definitions resource and each task's pinned definition, with the
digest recomputed so resumed tasks still match.
2026-10-25 17:31:16 +11:00
sam 340223de2e feat(modelloop): run a step's tool calls in parallel as each tool's classification allows 2026-10-25 16:45:34 +11:00
sam aa1e2ca660 feat(kit): central tools for harnesses through the egress proxy and gateway task-tool RPCs 2026-10-17 23:42:37 +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 a3e0e5c1bb fix(daemon): bound a package workflow's tools by its accepted authority 2026-10-11 03:48:31 +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 0a47a95cc3 fix(agentexec): enforce capture tool result schemas 2026-10-08 22:30:12 +11:00
sam c654ba7cd1 fix(modelloop): mark reasoning models from the carried model limits, in the gateway and the agent worker 2026-10-06 00:57:20 +11:00
sam 9f3ad773cd fix(modelloop): feed the cached model catalog to the runtime so reasoning models get the right token parameter 2026-10-06 00:45:09 +11:00
sam 84050080a7 fix(archiegateway,agentexec,agentworker): rely on ai-sdk reasoning token selection
Bump github.com/samcharles93/ai-sdk v0.1.40 -> v0.1.41. ai-sdk now sends
max_completion_tokens for reasoning models and omits temperature when the
model rejects it, so the max_tokens workarounds are no longer needed: the
chat-turn bound is set directly and MCP sampling loses its reasoning
branch. Classic models keep their existing bounds.
2026-10-05 22:56:48 +11:00
sam 2ae46582a8 refactor(skill): delete Yaegi skill scripts and the run_go_script tool
No shipped skill used plugins/. The agent request drops its Plugins field with
the staging code behind it.
2026-10-04 03:07:39 +11:00
sam 1e435e9224 style: adopt task fmt output; regenerate contracts and sqlc code; gateway cleanup derives its context
goimports during the service refactors rewrote generated files; they are
restored from their generators.
2026-10-04 01:48:32 +10:00
sam 07d4fcb627 refactor(agentexec): model loop moves to agentexec/modelloop
agentexec's wire contract, harness adapters and log forwarding no longer
import the AI SDK; the in-process model loop, runtime, tool set and sampling
helpers that do now live in agentexec/modelloop, used by the Gateway and
archie-agent only.
2026-10-04 01:36:13 +10:00
sam bf33ec9535 refactor(agentrun): agent stage contract moves to the domain layer
internal/domain/workflow imported internal/agentexec for its protocol types,
pulling the AI runtime into every binary that touches workflow types. The
request, result and runner contract now live in internal/domain/agentrun;
agentexec keeps the runtime. The model catalog cache moves from servicekit
into modelcatalog and WaitForPeer into staterpc, so archie-state-store and
archie-messaging no longer link the AI runtime.
2026-10-04 01:33:00 +10:00
sam 3360cddb5a refactor(staterpc,releaseupdate,agentexec,storecontract,worktree): cut comments down to what the code does 2026-10-03 22:00:43 +10:00
sam 030b4ed176 refactor: strip doc and issue narration from code comments
Remove PRD/doc paths, bead IDs, phase notes and deleted-test references
from comments; drop test-only hooks left by the test deletion. Includes
in-progress archied bootstrap, gateway and query changes.
2026-10-03 21:47:46 +10:00
sam 780bc44d82 refactor: testing from the beginning. a journey in 150k lines of tests to none 2026-10-03 21:16:25 +10:00
sam 7ca81dc85a feat(agentworker): answer MCP sampling from the task's model
A task hosts the same configured MCP servers as a chat turn, but the
worker's provider builder passed no sampling handler, so a task-scoped
server's sampling/createMessage request got a JSON-RPC method-not-found
while the identical request succeeded on the chat path. The worker runs
its own model runtime, so it can answer.

Thread a sampling handler built from the task's builder model into the
worker's MCP provider builder, beside the chat path's handler. Build the
task's model runtime once and share it with the stage runners instead of
creating a second client for sampling.

Move the sampling message mapping and max-token bound to agentexec, next
to the runtime both paths call models through, so a server gets the same
bound on either path. Carry the catalog's reasoning flag in
ModelLimits: a reasoning-class builder model must omit the max-token
bound its provider rejects, exactly as the chat path does.
2026-10-02 12:20:31 +10:00
sam ff3f80e4c1 fix(logging): file a task log record under its own event time
A remotely-shipped task log entry was re-emitted through the sink with
slog.Logger.LogAttrs, which stamps time.Now(): the entry's own event time
was discarded and replaced with the daemon's receipt time. Because a
container's records arrive in bursts, a multi-minute run collapsed onto a
few milliseconds of receipt timestamps and its timing became
unrecoverable -- which stage ran when, how long an iteration took, whether
a stage stalled.

Write now builds the record from entry.Time (falling back to receipt time
only for an entry that carries none) and dispatches it through the same
handler, preserving the sink's level gate. Receipt time stays available
separately in the record's received_at field, so event time and arrival
time cannot be confused. The agent-side producer already carried the
record's own time; a test now pins that half too.
2026-10-02 10:33:24 +10:00
sam 6bf24d73db test(agentexec): make seeded git repos independent of host signing config
The temp-repo fixtures in agentexec, domain/workflow and agentworker seeded
a repository and committed without disabling gpg signing, so a host with
commit.gpgsign=true and an agent-only SSH signing key failed those tests at
the fixture's own git call ("Couldn't get agent socket?").

Each producer now writes commit.gpgsign=false into the temporary repository's
own config, which also covers commits made inside the seeded repo by a
subprocess (the fake harness). A regression test per package pins the
fixture's hermeticity under a hostile global git config.
2026-10-02 03:06:48 +10:00
sam 1fa3acdbe9 fix(security): bound harness IDs to int and drop a summed map size hint
deploy / gate (push) Failing after 3m38s
deploy / build-and-push (push) Skipped
deploy / dist-zip (push) Skipped
Clears the five open CodeQL alerts. harnessCredential parsed uid and gid
as 32-bit and os.Chown converted them to int, which wraps on a 32-bit
build; they are now parsed to 31 bits, so an out-of-range id is refused
where it is read. mergeMapping sized its map by len(base)+len(override);
it now hints len(base) and lets the map grow.
2026-09-28 02:20:25 +10:00
sam cdc3e83740 chore(agentexec): strip the recording session's personal tool list from claude fixtures 2026-09-27 18:35:28 +10:00
sam 585df0f81e feat(agentexec): output adapters for Codex, Pi, OMP and Copilot CLI 2026-09-26 10:28:40 +10:00
sam a8c8dfbd7d feat(daemon): run a Kit profile's task in its Kit container on the egress path 2026-09-26 10:14:33 +10:00
sam 0d042b72d9 feat(agentexec): run a Kit task's stages on its harness through a named output adapter 2026-09-26 09:20:30 +10:00
sam 7a54db0cc6 fix(agentexec): count Claude cache tokens as prompt tokens, as the built-in loop does
Claude's input_tokens excludes cache reads and writes; the built-in
loop's Anthropic provider folds them into PromptTokens so budgets count
total prompt size. The harness adapter now does the same, so one budget
means the same on either runner.
2026-09-26 09:11:57 +10:00
sam 4067949ce8 feat(agentexec): parse Claude Code's stream-json output for the harness runner
Pairs tool_use and tool_result blocks by ID into tool-call reports, keeps
the latest session_id, takes the run's usage from the result event and
falls back to per-message usage, counted once per message ID, for a run
killed before its result. Tested on streams recorded from Claude Code
2.1.282 and scrubbed of host detail.
2026-09-26 09:11:57 +10:00
sam 235b9c1a87 refactor(agentexec): own the capture file format in agentexec, not the app layer 2026-09-26 03:30:19 +10:00
sam 37d913c5fd feat(agentexec): serve capture tools to harnesses over MCP
archie-agent mcp is a stdio MCP server, on the official go-sdk, serving a
stage's capture tools and appending accepted calls to a captures file.
The harness runner writes the tool spec and a standard mcpServers config,
registers it through HarnessSpec.MCPConfig, and reads the captures back
into the stage result. The captures file is writable by the harness user,
so the worker validates every record again with the same acceptCapture
the built-in loop now shares, and forged, undeclared or over-cap records
are dropped. A stage with capture tools on a harness without MCP is
refused before it runs.
2026-09-26 03:28:58 +10:00
sam 5b6572bcea feat(agentexec): run harness invocations as the Kit's user with no inherited environment
A Kit task runs worker and harness in one container, separated by user:
the worker keeps its credentials as root, and every harness invocation
runs as the Kit's user with only its declared environment. A nil env
would inherit the worker's, so the runner always passes an explicit one;
root is refused as a harness user, and a root worker refuses to start a
harness with no user. The PRD's container layout now says one container
per run, with the relay forwarding the proxy, NATS and the State Store.
2026-09-26 03:08:34 +10:00
sam e7611903a8 feat(agentexec): harness runner driving a Kit's agent headlessly
A stage whose request carries a HarnessSpec runs the Kit's CLI as a
subprocess in its own process group, with the mission and archie's rules
as the prompt. After every invocation the runner compares HEAD, every ref
and the git configuration, and the worktree against a pre-run baseline:
any ref or config change, a protected-path edit, or a write in a
read-only stage parks the step whatever the CLI claimed. The runner runs
the gate itself and resumes the same session with the gate's output, up
to the gate's failure cap. The wall clock and cancellation kill the whole
process group, so the CLI's children cannot outlive the step.
2026-09-26 02:54:03 +10:00
sam d8bec1ae06 fix(eda): clear completed tasks; document the tool allowlist's reach and unchecked profile names
ClearTerminalTasks now deletes completed tasks, with a test that holds its
SQL list to taskstate.Terminal. The AllowTools comment says agentloop's
built-in file tools are outside the allowlist and why. The PRD model
diagram lists only what a profile holds (image, tools), and the profile
field and AgentProfile say a workflow's profile name is resolved at
dispatch, not when the workflow is saved.
2026-09-24 20:53:51 +10:00
sam d320a1a01a feat(eda): workflow inputs, optional repository and agent profiles
Workflow definitions declare inputs, repository (none, optional or
required; default required) and profile. A binding assigns each input
from a mapped parameter or a constant and may take its repository from a
parameter, which must name a configured repo; both are checked when the
binding is saved and again at dispatch. Inputs travel as structured task
data (tasks.inputs, the task brief, and a JSON block in the agent's
mission), never as body text.

A task with no repository runs in a per-task scratch workspace with no
clone and no publication grant. The agent.run step runs one mission
without a repository and finishes the task in the new completed status.

Agent profiles live under [containers.profiles.<name>]: an image and an
allowlist over the tools archie adds (MCP, scripts, plugins).
[containers].image is the default profile; an unconfigured profile parks
the task. The workflow is now pinned before the container starts so the
profile can choose its image.

Also stop two archied tests writing embedded NATS state into the package
directory, refuse an empty state_dir for embedded NATS, and drop the NATS
store files an earlier lane commit tracked.
2026-09-24 20:29:33 +10:00
sam c94986cf38 Merge pull request #878 from samcharles93/feat/attempt-attribution
Feat/attempt attribution
2026-09-19 02:48:25 +10:00
sam fe7ee44dcc fix(workflow): require a workflow from triage when a code change is needed
The selection criteria alone left workflow optional and defaulted an omitted
or unrecognised value to implement, so a decide call could still skip the
decision and send an unsettled capability straight to the builder.

Omission is now a retryable rejection: CaptureTool gains RequiredWhenTrue, a
conditional requirement a flat RequiredFields entry cannot express, because
the close-without-change branch has no workflow to name and must not be made
to invent one. A rejected call is not recorded, so it does not consume
MaxCalls and the model retries with the field.

A value outside the enum now fails the stage rather than degrading to
implement: reaching OnResult with one means the model ignored its own schema.

Splits validateCaptureArgs into per-constraint checks (cyclop) and carries
the regenerated contracts.json.
2026-09-19 00:33:39 +10:00
sam 2c09f42ff4 feat(tasks): attribute events, changes and config to the attempt that produced them
The events table had no attempt column, so two attempts of the same task were
indistinguishable: a retried task's stage events interleaved into one timeline and
nothing could say which run produced a given result.

- events gains an `attempt` column through the existing presence-gated
  migrateTasks loop rather than a version bump, so user_version stays 3 and an
  existing archie.db is migrated in place. A legacy-database test builds the
  pre-change schema and proves both that existing rows survive unattributed
  (attempt = 0) and that a new value round-trips.
- Event.attempt = 11 and ReadTaskLogRequest.stage = 9 are field additions to
  existing messages: no new RPC, no new table, and no change to any canonical error
  message that crosses the state-store wire.
- Every task event the workflow engine emits now carries the attempt. Producers
  that have no attempt yet (task_queued, binding_dispatch_failure) stay
  unattributed rather than guessing one.
- Changed files are captured as a durable changes_captured event at commit and push
  time, reusing the existing three-dot diff primitive: paths from FilePatches(),
  counts from Stats(), base_sha recorded as the merge base actually diffed rather
  than the base tip, a 200-file cap with totals kept complete. Consumption goes
  through an unexported optional interface so workflow.Trees -- the yaegi capability
  surface for interpreted stage plugins -- is not widened. A capture is also taken
  once the pull request exists, so the capture can link it.
- The per-attempt configuration is captured as a durable config_captured event
  carrying the non-secret config.TaskConfig, instead of a new table plus two new
  state-store RPCs: events are already this system's per-attempt provenance stream,
  so a table would have given one feature two provenance mechanisms. Redaction is
  proven against canary secrets. Measured payload: 539 bytes typical, 2333 bytes at
  20 models.
- Captures carry a schema marker and are refused unless their key set matches the
  declared payload shape, so drift surfaces as "recorded but could not be read"
  rather than as a capture of zeroes.
- Three read endpoints serve the attempt rail, the changed-file captures and a raw
  debug view, mapping to UI-owned response types rather than dumping store types.
- The task-log read gains a stage filter, bound to the stage logger for that
  stage's own lines; agent and tool output records no stage.

A task-scoped grant may already insert events on its own task, so this provenance
is exactly as trustworthy as the rest of that task's event stream. That limit is
documented rather than papered over.
2026-09-18 18:32:59 +10:00
sam 33da682887 refactor(lint): extract more helpers to clear nestif/gocyclo findings
- toolset.go: extract requireApproval out of toolExecute (nestif)
- gateway.go: extract handleDeny/handleVersion/handleUpdateCheck/
  handleRestartAdapter out of dispatchLocal (gocyclo 26 -> under 20)
- cronstore/store.go: extract applyPatch/applyScheduleChange out of
  Update (nestif, gocognit)

Behavior-preserving.
2026-08-31 21:28:46 +10:00