Compare commits

...
75 Commits
Author SHA1 Message Date
sam 70251ff3ce chore(release): v1.61.0
Quality / check (push) Successful in 7m16s
deploy / gate (push) Successful in 4m44s
deploy / archied-image (push) Successful in 3m0s
deploy / agent-image (push) Successful in 3m35s
deploy / dist-zip (push) Successful in 2m39s
2026-11-08 04:59:28 +11:00
sam 76dc34a17c fix(distribution): trust the rotated catalogue signing key (#1790)
Quality / check (push) Canceled after 5m9s
deploy / archied-image (push) Successful in 2m58s
deploy / dist-zip (push) Skipped
deploy / gate (push) Successful in 4m32s
deploy / agent-image (push) Successful in 12s
2026-11-08 04:54:32 +11:00
sam a6a0c7c11e fix(archied): say why the canary can't compare migrations
deploy / gate (push) Successful in 4m41s
deploy / archied-image (push) Successful in 2m57s
deploy / agent-image (push) Successful in 11s
deploy / dist-zip (push) Skipped
Quality / check (push) Successful in 6m27s
2026-11-08 03:05:29 +11:00
sam f759387eb7 chore: ignore the fleet run directory 2026-11-08 02:48:29 +11:00
sam 21b3591da1 fix(e2e+ui): e2e checks for the audit's failure classes, and the gaps they caught (#1767)
deploy / gate (push) Successful in 4m48s
deploy / dist-zip (push) Skipped
Quality / check (push) Successful in 6m11s
deploy / archied-image (push) Successful in 3m31s
deploy / agent-image (push) Successful in 12s
loading-errors now forces each list API (/api/tasks, /api/captures,
/api/event-types, /api/tokens, /api/access/policies) to fail and asserts
the page shows its error surface, never its empty-state copy.
accessibility gains two static-route sweeps: every input/textarea/select/
button must have an accessible name, and every keyboard focus stop's
indicator (ring var, outline, shadow or focus border) must clear 3:1
against the background behind it. phone-overflow sweeps every static
route at 390x844 for horizontal scroll. The sweeps caught real gaps,
fixed here: the DurationInput amount input had no accessible name and
its unit select no focus indicator, and TabsContent and CodeBlockContent
were focusable with outline-none and no ring. The stale 'Sections' list
selector now matches the nav landmark.
2026-11-07 17:08:10 +11:00
sam e9937ded6b fix(ui): a failed answer keeps the question card and its filled-in answer (#1728)
Quality / check (push) Successful in 9m2s
deploy / gate (push) Successful in 4m49s
deploy / archied-image (push) Successful in 3m25s
deploy / agent-image (push) Successful in 11s
deploy / dist-zip (push) Skipped
answerAsk now clears the pending question only after api.chatAnswer
accepts it, and resolves whether it did. On failure the new askError ref
carries the message, shown on the card as a ReadError whose Retry resends
the last answer; ChatAskCard keeps the typed reply until acceptance and
routes every choice button through the same path, and the questionnaire
submits through the card so its answers are retryable too. askError resets
on turn start, on a new ask, and on conversation switch.
2026-11-07 01:23:10 +11:00
sam bd1acdd0cc fix(ui): switching conversation detaches the running turn (#1729)
Quality / check (push) Canceled after 23s
deploy / archied-image (push) Successful in 3m24s
deploy / agent-image (push) Successful in 11s
deploy / dist-zip (push) Skipped
deploy / gate (push) Successful in 4m47s
selectSession now bumps a session epoch and aborts the in-flight stream
before switching, and the transcript read is applied only if the session is
still current. sendMessage captures the epoch at turn start: a detached turn
stops reading, cancels the reader, and writes no messages, streamingTurn or
session id; only the turn still owning the abort controller clears isSending,
so it cannot unlock the composer for a turn running in the new conversation.
2026-11-07 01:20:49 +11:00
sam a2c6341544 fix(ui): policy form opens blank on New, id read-only when editing, dirty draft asks first (#1730)
Quality / check (push) Canceled after 4m48s
deploy / gate (push) Successful in 4m52s
deploy / archied-image (push) Successful in 3m26s
deploy / dist-zip (push) Skipped
deploy / agent-image (push) Successful in 11s
New policy now resets the draft instead of reusing the open policy's, so
Save cannot overwrite the policy that happens to be open. The id input is
readonly while editing a persisted policy, so a rename can no longer write
a second policy. New/Edit/Cancel route through openForm, which asks for
confirmation before discarding a dirty draft.
2026-11-07 01:16:48 +11:00
sam cb3372ac76 fix(ui): terminal palette is theme tokens, switches live (#1739)
Quality / check (push) Canceled after 3m23s
deploy / archied-image (push) Successful in 3m22s
deploy / agent-image (push) Successful in 11s
deploy / dist-zip (push) Skipped
deploy / gate (push) Successful in 4m51s
--terminal-bg/fg/cursor/selection are defined per theme in style.css
(dark keeps the deep well, light gets a paper inset) and exposed as
bg-terminal. Both xterm surfaces (SetupTerminal, SandboxView) read them
via lib/terminalTheme and re-apply on every data-theme flip, so the
theme switch is live; the raw #0b0f14/#e6edf3/#264f78 hex is gone.
2026-11-07 01:13:41 +11:00
sam 48e605a545 fix(ui): dead tokens replaced or deleted — text-ok, --radius-lg, real surface (#1740)
Quality / check (push) Successful in 6m33s
deploy / gate (push) Successful in 4m49s
deploy / archied-image (push) Successful in 3m22s
deploy / dist-zip (push) Skipped
deploy / agent-image (push) Successful in 11s
text-success was not a token (now text-ok) in ChatToolCall and BindEvent;
var(--radius) is undefined (now var(--radius-lg)) in input-group and Sonner;
Sonner's hsl(var(--popover) / 0.9) wrapped a hex (now a plain hex); the
undefined cn-font-heading class is deleted from the nine title components.
2026-11-07 01:06:42 +11:00
sam 9a4e59f0e0 fix(ui): tab, toggle, input and idle-text contrast meet their targets (#1741)
Quality / check (push) Canceled after 3m37s
deploy / archied-image (push) Successful in 3m24s
deploy / agent-image (push) Successful in 11s
deploy / dist-zip (push) Skipped
deploy / gate (push) Successful in 4m48s
Inactive tab labels use text-fg-muted instead of foreground/60 (was
4.1:1). The selected toggle/toggle-group segment gains a sage-deep ring
(5.3:1 light, 3.9:1 dark) instead of the near-invisible muted fill.
--input rises to 3.6:1 (light) and 3.5:1 (dark) on the card; dark --idle
rises to 5.1:1 on --accent. Values computed against the style.css tokens.
2026-11-07 01:03:19 +11:00
sam 820beda72f fix(ui): empty states wait for the first read to settle (#1742)
Quality / check (push) Successful in 6m42s
deploy / gate (push) Successful in 4m49s
deploy / archied-image (push) Successful in 3m27s
deploy / dist-zip (push) Skipped
deploy / agent-image (push) Successful in 11s
Skills, curators, QuickTask's workflow list, the alerts bell and the
extensions catalogue now show a loading state until the first read
settles, then the data, the error, or the empty state. Draft-backed lists
(agent profiles, playbooks, credential bindings) show their empty state
only once the draft exists, not beside the load error.
2026-11-07 00:54:25 +11:00
sam c6f70de82d fix(ui): settings pages show a skeleton or the error until their draft loads (#1743)
Quality / check (push) Successful in 7m32s
deploy / agent-image (push) Successful in 11s
deploy / gate (push) Successful in 4m59s
deploy / archied-image (push) Successful in 3m27s
deploy / dist-zip (push) Skipped
New settings/KindGate renders the store's per-kind loading/error state
where the form would be: a skeleton while loading, ReadError with Retry
on a failed read. Wired into Tools, SystemModels, SystemRepos,
SchedulingPolicy, ContainerRuntime, OrgAlerts, Instructions, Plugins,
SystemTasks, UpdateSettings, IdentityGrants, OrgModels, Channels and
Authentication pages, which no longer render blank until the draft
exists.
2026-11-07 00:42:28 +11:00
sam f37f6e2453 fix(ui): one error presentation with Retry across the dashboard (#1744)
Quality / check (push) Successful in 6m56s
deploy / gate (push) Successful in 4m58s
deploy / agent-image (push) Successful in 11s
deploy / dist-zip (push) Skipped
deploy / archied-image (push) Successful in 3m28s
New components/ui ReadError (destructive Alert + message + optional Retry)
is now the single read-error presentation, used by Versions, Services,
Canary, OrgAccess, Denials, Tokens, Audit, Extensions, Harness and the
Status page (one page-level error instead of one per card). PanelError is
the same alert for task panels, and TasksState's error kind, the attempts
panel and the task-log pane all offer Retry. Org policy save errors sit
beside the Save button instead of at the page top.
2026-11-07 00:33:49 +11:00
sam f7a999e317 fix(ui): failed task pages keep rows with an inline error; refreshes don't flash (#1745)
deploy / archied-image (push) Successful in 3m28s
deploy / agent-image (push) Successful in 11s
deploy / dist-zip (push) Skipped
Quality / check (push) Successful in 6m31s
deploy / gate (push) Successful in 4m49s
A failed Load more or live refresh sets moreError beside the Load more
button instead of replacing the table; only a first-load failure takes
the table down. The no-match empty state says search covers loaded rows
when more pages exist. refreshAll keeps attempts/events (keep=true) and a
failed task-list read keeps the last known list, so panels no longer
flash to loading on every live update.
2026-11-07 00:25:36 +11:00
sam 45494d5b43 fix(ui): logs page error no longer flickers, live entries pruned, unavailable clears (#1746)
deploy / gate (push) Successful in 4m51s
deploy / agent-image (push) Successful in 11s
deploy / dist-zip (push) Skipped
Quality / check (push) Canceled after 2m59s
deploy / archied-image (push) Successful in 3m28s
readError is cleared only by a successful read instead of at the start of
every 5 s refresh, so an outage shows a steady error beside the stale
lines. liveEntries is pruned past 1500 (the list shows 1000) so the map
cannot grow unbounded. streamState leaves "unavailable" as soon as live
entries flow again.
2026-11-07 00:22:40 +11:00
sam 9ca9545847 fix(ui): alert state is shared; a failed dismiss keeps the alert and says why (#1747)
Quality / check (push) Canceled after 3m8s
deploy / gate (push) Successful in 4m51s
deploy / archied-image (push) Successful in 3m30s
deploy / agent-image (push) Successful in 12s
deploy / dist-zip (push) Skipped
use-alerts.ts now holds the alerts/failed refs at module scope (wired once),
so the bell and the dashboard card see the same list. dismiss removes the
alert only after the server confirms; on failure it stays and the reason
shows in the bell. dismissAll reports any rejection and reloads.
2026-11-07 00:19:31 +11:00
sam fa14e7da80 fix(ui): chat failures are shown where they happen and leave the real state (#1748)
deploy / gate (push) Successful in 4m47s
deploy / agent-image (push) Successful in 11s
Quality / check (push) Canceled after 1m18s
deploy / archied-image (push) Successful in 3m27s
deploy / dist-zip (push) Skipped
Mic permission failure shows an inline alert; setModel only switches on
server confirmation and reverts + reports in the transcript on failure;
a failed session refresh after a turn no longer becomes a retryable turn
error; a voice clip during a turn queues instead of dropping; blocked
localStorage falls back to a per-load channel id; a failed install offers
Retry/Not now again; rename/branch use a labelled Dialog instead of
window.prompt; a failed cancel is reported; a stopped turn no longer
blocks transcript reloads.
2026-11-07 00:17:44 +11:00
sam 12eea5eee3 fix(ui): destructive actions confirm via AlertDialog, naming the item (#1749)
deploy / gate (push) Successful in 4m49s
deploy / archied-image (push) Successful in 3m29s
deploy / dist-zip (push) Skipped
Quality / check (push) Canceled after 7m10s
deploy / agent-image (push) Successful in 12s
Delete policy, remove member, revoke token, remove passkey and delete
source now open the shared AlertDialog confirmation that names the item,
matching the binding/mapping/event-type deletes that already confirmed.
2026-11-07 00:11:16 +11:00
sam ef623f6c68 fix(ui): rows keep table semantics; first-cell link/button is the focus target (#1750)
deploy / archied-image (push) Successful in 3m29s
deploy / agent-image (push) Successful in 11s
Quality / check (push) Successful in 6m40s
deploy / gate (push) Successful in 4m59s
deploy / dist-zip (push) Skipped
DataTable, ActivityRow, LiveActivityCard's group row and CaptureRow no
longer put role=link/button, tabindex or aria-label on the <tr>. The row
keeps its mouse click handler; the first cell carries a real RouterLink
(DataTable rowLink) or button with the accessible name, so keyboard and
screen-reader users get a focus target and rows keep row/cell semantics.
The live-activity group row's disclosure button and open button are now
siblings, not nested.
2026-11-07 00:00:41 +11:00
sam ee2d19d011 fix(ui): real heading structure and a nav landmark for the sidebar (#1751)
Quality / check (push) Canceled after 4m2s
deploy / archied-image (push) Successful in 3m30s
deploy / agent-image (push) Successful in 11s
deploy / gate (push) Successful in 4m52s
deploy / dist-zip (push) Skipped
CardTitle and EmptyTitle take an as prop (default h2) and render through
component :is, so card and empty-state titles are headings; Section passes
as="h2" under the page's h1. The sidebar menu list now sits in
<nav aria-label="Sections">.
2026-11-06 23:56:34 +11:00
sam 53c6b3e413 fix(ui): disabled reasons and truncated values reachable without hover (#1752)
Quality / check (push) Canceled after 4m7s
deploy / gate (push) Successful in 4m49s
deploy / archied-image (push) Successful in 3m32s
deploy / agent-image (push) Successful in 12s
deploy / dist-zip (push) Skipped
Disabled New run and task-row buttons now carry their reason in a Tooltip
on a focusable wrapper (disabled buttons cannot take focus), and New run
says when the task list has not loaded. Truncated values (binding filter
and failure reason, activity rows, live alerts, event-type rule summaries,
source URLs, mapping field paths) are focusable spans whose full text is in
a Tooltip. The org switcher has distinct loading ('…' tile, 'Loading…')
and failed ('!' tile, 'Organisations unavailable') states.
2026-11-06 23:52:25 +11:00
sam d58125c0bc fix(ui): combobox pattern for the slash palette, named palette input, bound labels (#1753)
Quality / check (push) Canceled after 4m45s
deploy / archied-image (push) Successful in 3m28s
deploy / agent-image (push) Successful in 12s
deploy / dist-zip (push) Skipped
deploy / gate (push) Successful in 4m48s
The composer textarea is now role=combobox with aria-expanded,
aria-controls and aria-activedescendant wired to the slash palette's
listbox ids. CommandInput has an accessible name. FieldLabels without a
control beside them are bound: for/id on selects, inputs, switches and
duration inputs across channels, curators and schedules pages, and
role=group + aria-labelledby/aria-label on the TagsInput, secret-ref and
rule row groups.
2026-11-06 23:47:41 +11:00
sam 029198ff61 fix(ui): generation-guard the remaining async loads and polls (#1754)
Quality / check (push) Canceled after 7m38s
deploy / gate (push) Successful in 4m51s
deploy / archied-image (push) Successful in 3m26s
deploy / agent-image (push) Successful in 12s
deploy / dist-zip (push) Skipped
Schedules, command palette, skills, repos, services, secret engines,
captures, event types, curators, sources, channels, mappings (list and
preview), dispatch links and workflow history now capture a generation
at request start and drop their writes unless still latest — the guard
logs/state.ts already used. A failed schedules poll keeps the last good
rows beside the inline error instead of replacing the table, and a
failed skills/secret-engines poll no longer clears the list.
2026-11-06 23:40:03 +11:00
sam 0b08fc5242 fix(ui): harness setup terminal reports the real failure and can be reopened (#1755)
Quality / check (push) Canceled after 5m18s
deploy / gate (push) Successful in 4m48s
deploy / archied-image (push) Successful in 3m29s
deploy / agent-image (push) Successful in 11s
deploy / dist-zip (push) Skipped
A websocket that never opened stays "failed" — onclose no longer
overwrites it with "closed", so the terminal says "Could not open the
setup terminal." instead of "The terminal session ended.". The terminal
emits ended when its session closes, and HarnessPage resets terminalOpen
on it (plus a Close terminal button), so reopening starts a fresh session
instead of requiring a page reload. The harness load failure now has a
Retry button.
2026-11-06 23:34:38 +11:00
sam 93609c7ab3 fix(ui): eleven small dashboard state bugs from the audit (#1756)
Quality / check (push) Canceled after 3m13s
deploy / agent-image (push) Successful in 11s
deploy / gate (push) Successful in 4m50s
deploy / archied-image (push) Successful in 3m28s
deploy / dist-zip (push) Skipped
Install disables when no installable snapshot; retention hours is empty
until loaded instead of showing 0; the audit Action field carries all
comma-separated actions; saving a view over an existing name asks first;
"Password changed" is a notice, not a destructive alert; "Sign in
again" appears only on an expired session (classifyActionError) on
account, authentication and create-account; sign-out failure sets the
store error instead of an unhandled rejection; the workflow identity
select shows the seeded agent without writing it into the YAML;
unparseable durations show an inline error instead of snapping back;
CardAccordion keys items by id; curator add() explains a no-op.
2026-11-06 23:31:47 +11:00
sam 461d3dc1f7 fix(ui): phone-width overflows in the audit's flagged rows, with an e2e check (#1757)
Quality / check (push) Successful in 6m36s
deploy / agent-image (push) Successful in 11s
deploy / dist-zip (push) Skipped
deploy / gate (push) Successful in 4m54s
deploy / archied-image (push) Successful in 3m23s
Grants rows and toggle groups wrap with a smaller phone label; the token
form wraps and its input flexes; the sources name column caps at full
width and its action buttons wrap; the rule-editor condition row wraps
with a full-width phone select; the payload tree indents less and breaks
long JSON leaves; the topbar instance badge truncates; org agent rows
keep Assign inline and move Rename/Suspend/Retire into a MoreMenu. New
e2e check phone-overflow.cjs visits these views at 390x844 with seeded
data and asserts no horizontal scroll.
2026-11-06 23:23:16 +11:00
sam 94f79177a2 fix(ui): adopt the type scale across components/ui and pages (#1758)
deploy / gate (push) Successful in 4m53s
Quality / check (push) Successful in 6m42s
deploy / archied-image (push) Successful in 3m20s
deploy / dist-zip (push) Skipped
deploy / agent-image (push) Successful in 11s
components/ui and the pages now use only the style.css scale:
text-[0.8rem]→text-label, text-[0.625rem]→text-micro, text-base/text-sm→
text-label (rows and controls), text-xs→text-caption (captions and
secondary text), text-lg/text-xl/text-2xl→text-title (panel and page
titles). Rendered markdown (ChatMarkdown/ChatInline/markdown.ts) keeps
Tailwind steps as the issue allows. 196 files, class-token swaps only.
2026-11-06 23:07:46 +11:00
sam 9169e9db09 fix(ui): hand-rolled cards, headings, captions and empties use components/ui (#1759)
deploy / gate (push) Successful in 4m50s
Quality / check (push) Canceled after 4m48s
deploy / archied-image (push) Successful in 3m24s
deploy / agent-image (push) Successful in 11s
deploy / dist-zip (push) Skipped
Captions move from text-xs to the text-caption type-scale token across
the workflow/bind/schedule editors and curator card. The task page title
uses PageHeader's h1 classes. Hand-rolled boxes become Card (token
reveal, recovery code, pending extension update) or CodeView (install
progress); the startup-settings key/value table becomes MetadataList;
the harness empty row becomes TableEmpty; 'No tokens/dispatches/inputs/
removed' states become Empty; the step-editor problem line becomes a
destructive Alert. The risky-row danger tint in ToolsPage and
SystemReposPage now uses SettingRow tone="danger" instead of a
hand-written before: overlay (DANGER_ROW deleted).
2026-11-06 23:03:10 +11:00
sam 0d0cd09124 fix(ui): known lists use Select/TagsInput fed by the server catalogue (#1760)
deploy / archied-image (push) Successful in 3m17s
Quality / check (push) Successful in 8m37s
deploy / gate (push) Successful in 4m36s
deploy / agent-image (push) Successful in 12s
deploy / dist-zip (push) Skipped
Token lifetimes use Select instead of NativeSelect. Credential bindings
pick service from the store's credentialServices and organisation from
the org list (empty = default, via a sentinel). Curator and agent-profile
tools are TagsInput instead of comma text; curator model is a Select of
the configured providers' models from /api/config's catalog. Org and
system model pages pick provider class from the catalog's classes, and
SystemModelsPage's local CLASSES list is deleted. Chat model choice is a
Select of the server-served model list instead of a button row.

Left as free text because no server source exists: policy object kind,
curator memory engine, update channel labels, connection type default,
container default, step titles (x-step-types carries no titles).
2026-11-06 22:51:21 +11:00
sam 0dc5bf9a9b fix(ui): times honour the Appearance zone, one duration formatter everywhere (#1761)
Quality / check (push) Successful in 6m44s
deploy / archied-image (push) Successful in 3m44s
deploy / agent-image (push) Successful in 11s
deploy / dist-zip (push) Skipped
deploy / gate (push) Successful in 4m54s
lib/format.ts gains shortTime(value, zone) and becomes the home of the
single duration() formatter (moved from timeline-event.ts, which
re-exports it so the stage rail and duration.test.ts are unchanged).
LogTable's time column, audit's From/To chips and chat session titles now
format in the zone chosen under Appearance instead of the browser's.
StepNode, schedule runDuration, labels formatMs and ChatToolCall all
delegate to duration(), removing the "12ms" vs "12 ms" vs "under 1 s"
split.
2026-11-06 22:21:45 +11:00
sam ca107f6977 fix(ui): failed draft save is retried, then shown — never dropped silently (#1731)
persistDraft retried only while draftState.saving was true, but
settings.replace clears saving in its finally before resolving false, so
the retry branch could never run and a failed putDraft vanished. It now
retries on a bounded schedule (5 attempts, 500ms apart) regardless of the
saving flag, and on final failure sets localError so the editor's alert
says the draft was not saved while the edit stays in the editor. The
alert also reads the draft store's own error (workflows.draftState.error)
instead of only the definitions store's.
2026-11-06 22:09:56 +11:00
sam 4dad1415e6 fix(ui): workflow reads keep their error instead of showing empty (#1732)
Triggers: a failed bindings/schedules read surfaces as an error with
Retry in TriggerSettings instead of "Manual runs only." Run stages: a
failed attempts read sets stagesError, shown with Retry above the canvas
instead of a bare canvas. Bind form: failed reads show an error with
Retry rather than empty pickers, and save refuses to mappingUpdate an
existing mapping unless the bindings list actually loaded. History: a
failed revisions read shows the error with Retry in the revision list,
and "No recorded changes" appears only after a successful read.
2026-11-06 22:06:25 +11:00
sam 5bf0a696c6 fix(ui): failed reads show an error with Retry, not an empty view (#1733)
Chat: a failed conversation load sets sessionError, shown in the
transcript with Retry instead of the empty invitation (it no longer goes
to statusText in the settings popover). Command palette: failed task or
workflow reads render a "Couldn't load … — Retry" item in that group.
Denials: the table no longer shows "No denials recorded" under an
error. Mapping form: failed captures/event-types reads show an error row
with Retry instead of "No captured events yet." or a silently empty
select. Channels: a failed status read keeps the last badges and shows
an error with Retry. Sources: read failures are separate from write
failures, shown with Retry, and a loaded-but-empty list gets an empty
state.
2026-11-06 22:01:03 +11:00
sam 7a8d948fcb fix(ui): drop stale async responses with a generation guard (#1734)
TasksPage load/loadMore, use-task-run keyedLoad, TaskLogs loadOlder and
EventsPage loadCounts now each capture a generation at request start and
drop their writes unless they are still the latest request, the same
guard logs/state.ts already uses. dropCaches also stales every in-flight
keyed read, so a 30s timeout's null can no longer overwrite a newer
successful response, and the task-log pager no longer writes a page
fetched under the old attempt or filter.
2026-11-06 21:52:55 +11:00
sam d90f7d48d0 fix(ui): step card header is a keyboard-operable button, Edit in its menu (#1735)
The step header in StepNode.vue was a div with @click and no role, so a
step could not be opened from the keyboard, and Edit existed only in the
right-click canvas menu. The header now renders as a real <button> (with
aria-expanded) whenever the step can toggle, so Tab/Enter/Space open it;
while renaming it stays a div so the rename input is not nested inside a
button. The actions menu trigger moved to an absolute overlay at the
card's top-right (same spot as before, with a matching spacer in the
header row) and gained an Edit item that toggles the step, matching the
canvas context menu.
2026-11-06 21:48:55 +11:00
sam c5a43f21aa fix(ui): accessible names on fields that had only placeholders (#1736)
Step editor fields (lead, compact rows, long/YAML textareas and the
compact boolean switch), the org policy editor (policy id, level,
workspace id, object kind/id, Cedar textarea), the binding constant
value input, the log search and the channel tags inputs now each carry
an aria-label. On TagsInput the aria-label moved from the root div,
where screen readers never reach it, onto the actual TagsInputInput.
2026-11-06 21:43:23 +11:00
sam d3fbb0842f fix(ui): focus rings at full strength for 3:1 contrast (#1737)
The shared components used focus-visible:ring-ring/50 (and the base
outline-ring/50), which measured 2.1:1 on light and 2.4:1 on dark --
under the 3:1 WCAG 2.4.11/1.4.11 requires. All ring-ring and outline-ring
tokens in ui/src now render at full strength (5.3:1 light, 5.7:1 dark),
matching the :focus-visible outline floor in style.css.
2026-11-06 21:39:37 +11:00
sam dafd426316 fix(ui): startup failure is a centred panel that retries on its own (#1738)
When /api/auth/status fails, App.vue now shows a centred card with the
logo, a plain 'Can't reach Archie' statement and the detail in a
destructive Alert (role=alert), and retries the read every 5s until it
succeeds instead of offering only a manual Retry link. errorMessage in
api.ts only reads text/plain bodies, so a proxy's HTML error page no
longer passes through as the detail; other content types fall back to
the status line.
2026-11-06 21:36:17 +11:00
sam bec374aa92 ci(quality): stop installing ripgrep and setting sysctl from the gate (#1710)
The check job now runs on the archie-quality self-hosted runner under an
unprivileged user, so the sudo apt-get install and sudo sysctl steps fail,
exactly as the deploy gate did before 3b372b4a. Both are host provisioning:
ripgrep is installed on the runner host and the userns restriction belongs in
sysctl.d. The namespace check stays, without sudo.
2026-11-06 21:30:33 +11:00
sam 85b21578d8 ci(release): name the missing DEPLOY_TOKEN before checkout, and list what a mirror needs
Quality / check (push) Successful in 3m46s
deploy / gate (push) Successful in 4m29s
deploy / archied-image (push) Successful in 2m53s
deploy / agent-image (push) Successful in 12s
deploy / dist-zip (push) Skipped
Refs #1788
2026-11-05 00:33:18 +11:00
sam 03c25e7214 docs(workflow): say repo.commit never runs the checkout's git hooks
Quality / check (push) Successful in 3m44s
deploy / gate (push) Successful in 4m45s
deploy / archied-image (push) Successful in 3m0s
deploy / agent-image (push) Successful in 12s
deploy / dist-zip (push) Skipped
Refs #1792
2026-11-05 00:26:20 +11:00
sam c30624a59e feat(workflow): a handoff starts a separate linked run of the next workflow (#1791)
Quality / check (push) Successful in 4m12s
deploy / archied-image (push) Successful in 3m13s
deploy / gate (push) Successful in 4m42s
deploy / agent-image (push) Successful in 11s
deploy / dist-zip (push) Skipped
2026-11-03 20:56:03 +11:00
sam 3cf025fb4a fix(worktree): mirror fetch follows moved tags and rewritten branches
Quality / check (push) Successful in 6m46s
deploy / agent-image (push) Successful in 12s
deploy / dist-zip (push) Skipped
deploy / gate (push) Successful in 4m35s
deploy / archied-image (push) Successful in 2m55s
2026-11-03 10:00:04 +11:00
sam 81b784876b fix(ui): live task updates refresh in place instead of flashing the loading state
deploy / agent-image (push) Successful in 11s
deploy / dist-zip (push) Skipped
Quality / check (push) Successful in 3m52s
deploy / gate (push) Successful in 4m33s
deploy / archied-image (push) Successful in 3m14s
2026-11-03 09:51:43 +11:00
sam 6c42f64bfa chore(release): v1.60.0
Quality / check (push) Canceled after 18s
deploy / gate (push) Successful in 4m39s
deploy / archied-image (push) Successful in 2m52s
deploy / agent-image (push) Successful in 3m21s
deploy / dist-zip (push) Successful in 1m47s
2026-11-03 09:51:30 +11:00
sam d74216c6b7 docs(changelog): curator issue tools (#1770)
Quality / check (push) Successful in 4m3s
deploy / gate (push) Successful in 4m42s
deploy / archied-image (push) Successful in 2m49s
deploy / agent-image (push) Successful in 11s
deploy / dist-zip (push) Skipped
2026-11-03 09:43:45 +11:00
sam 0b7ce448ab fix(e2e): checks follow the sidebar navigation; trim stray button and caption whitespace (#1789)
Quality / check (push) Successful in 3m52s
deploy / gate (push) Successful in 4m36s
deploy / archied-image (push) Successful in 3m16s
deploy / agent-image (push) Successful in 11s
deploy / dist-zip (push) Skipped
2026-11-03 09:35:19 +11:00
sam 2fb8c09a00 refactor(webui): extract the workflows module (#1778)
deploy / gate (push) Successful in 4m37s
deploy / archied-image (push) Successful in 2m51s
deploy / agent-image (push) Successful in 11s
Quality / check (push) Successful in 4m23s
deploy / dist-zip (push) Skipped
2026-11-03 08:39:40 +11:00
sam 0442f1849f refactor(webui): extract the audit module and its export guards (#1778)
deploy / archied-image (push) Successful in 2m52s
deploy / agent-image (push) Successful in 11s
Quality / check (push) Canceled after 1m17s
deploy / gate (push) Successful in 4m33s
deploy / dist-zip (push) Skipped
2026-11-03 08:38:22 +11:00
sam cf049797b0 refactor(webui): extract the schedules and alerts modules (#1778)
Quality / check (push) Canceled after 1m5s
deploy / gate (push) Successful in 4m33s
deploy / dist-zip (push) Skipped
deploy / archied-image (push) Successful in 2m48s
deploy / agent-image (push) Successful in 12s
2026-11-03 08:37:15 +11:00
sam 18e6fe6af6 refactor(webui): extract the chat module and its ask/file state (#1778)
deploy / gate (push) Successful in 4m35s
deploy / agent-image (push) Successful in 11s
deploy / dist-zip (push) Skipped
Quality / check (push) Canceled after 1m5s
deploy / archied-image (push) Successful in 2m48s
2026-11-03 08:36:09 +11:00
sam e8a84d3e41 refactor(webui): extract the control-plane and logs modules (#1778)
Quality / check (push) Canceled after 1m17s
deploy / gate (push) Successful in 4m31s
deploy / archied-image (push) Successful in 2m49s
deploy / agent-image (push) Successful in 11s
deploy / dist-zip (push) Skipped
2026-11-03 08:34:51 +11:00
sam c5f4541cbf refactor(webui): extract the access policies and denials module (#1778)
Quality / check (push) Canceled after 1m5s
deploy / gate (push) Successful in 4m35s
deploy / dist-zip (push) Skipped
deploy / archied-image (push) Successful in 2m49s
deploy / agent-image (push) Successful in 12s
2026-11-03 08:33:46 +11:00
sam 369e5038ca refactor(webui): extract the identities and tokens module (#1778)
deploy / gate (push) Successful in 4m34s
deploy / archied-image (push) Successful in 2m52s
deploy / agent-image (push) Successful in 10s
deploy / dist-zip (push) Skipped
Quality / check (push) Canceled after 1m9s
2026-11-03 08:32:44 +11:00
sam 66862a1bab refactor(webui): extract the orgs resource module (#1778)
Quality / check (push) Successful in 4m2s
deploy / dist-zip (push) Skipped
deploy / gate (push) Successful in 4m39s
deploy / archied-image (push) Successful in 2m51s
deploy / agent-image (push) Successful in 11s
2026-11-03 08:28:36 +11:00
sam 25dcdb2d19 refactor(webui): extract the extensions resource module (#1778)
Quality / check (push) Successful in 5m1s
deploy / gate (push) Successful in 4m35s
deploy / archied-image (push) Successful in 2m46s
deploy / agent-image (push) Successful in 10s
deploy / dist-zip (push) Skipped
2026-11-03 08:22:02 +11:00
sam 1830a89898 refactor(webui): extract the harness resource module (#1778)
deploy / gate (push) Successful in 4m34s
deploy / archied-image (push) Successful in 2m55s
deploy / agent-image (push) Successful in 12s
Quality / check (push) Successful in 3m56s
deploy / dist-zip (push) Skipped
2026-11-03 08:17:12 +11:00
sam 6e718e165a feat(curator): issue_list and issue_create tools via the daemon (#1770)
Quality / check (push) Successful in 4m6s
deploy / gate (push) Successful in 4m30s
deploy / archied-image (push) Successful in 2m50s
deploy / agent-image (push) Successful in 12s
deploy / dist-zip (push) Skipped
2026-11-03 08:12:05 +11:00
sam 0a37b551f5 refactor(webui): extract the tasks resource module (#1778)
Quality / check (push) Successful in 3m47s
deploy / gate (push) Successful in 4m32s
deploy / archied-image (push) Successful in 2m51s
deploy / agent-image (push) Successful in 10s
deploy / dist-zip (push) Skipped
2026-11-03 08:06:56 +11:00
sam 864b87faa6 refactor(webui): extract event-types, mappings and sources modules (#1778)
Quality / check (push) Successful in 3m52s
deploy / gate (push) Successful in 4m29s
deploy / archied-image (push) Successful in 2m48s
deploy / agent-image (push) Successful in 11s
deploy / dist-zip (push) Skipped
Same pattern as the bindings pilot: eventTypesAPI (registers unconditionally,
list answers enabled:false without a store), mappingsAPI and sourcesAPI
(constructed only when their store is configured, 503 stub otherwise).
Shared helpers unsignedSources/recentRefusals/listedIn stay on Server; the
dispatch-ledger handler stays shared for captures/tasks. task check green.
2026-11-03 07:42:50 +11:00
sam d55d8580cf refactor(webui): extract the bindings resource module (#1778)
Pilot for one-module-per-resource: bindingsAPI owns the /api/bindings*
routes and their handlers, constructed only when the binding store is
configured; a 503 stub mounts the same routes otherwise, preserving the
old not-configured answers. Handlers move off *Server unchanged in
behavior; the shared dispatch-ledger handler stays on Server for the
captures/tasks routes. task check green.
2026-11-03 07:01:12 +11:00
sam 63229d1711 refactor(storecontract): assert every store interface at its implementation (#1781)
Complete the var _ storecontract.X = (*T)(nil) sweep so each of the 31
contract interfaces is compiler-checked against at least one adapter:
postgres.Store (19), postgres.EDA (11, consolidated into edastore.go),
staterpc.Client (29). Removes the scattered single-line assertions from
harness_secrets.go, edastore_eventtypes.go, sources.go, run_credentials.go.
task check green.
2026-11-03 06:24:09 +11:00
sam 56044476fc chore(release): v1.59.0
deploy / gate (push) Successful in 4m21s
deploy / archied-image (push) Successful in 2m45s
deploy / agent-image (push) Successful in 3m15s
deploy / dist-zip (push) Successful in 2m35s
Quality / check (push) Successful in 3m53s
2026-11-03 06:15:09 +11:00
sam 796bfd0045 docs(changelog): forge extensions gain CreateIssue
deploy / gate (push) Successful in 5m16s
deploy / archied-image (push) Successful in 3m2s
deploy / agent-image (push) Successful in 11s
deploy / dist-zip (push) Skipped
Quality / check (push) Successful in 7m12s
2026-11-03 02:43:25 +11:00
sam abcb72a76a feat(curator): dry-run self-intake pass with run inspection tools (#1770)
Curators could only reach the workflow/integration tools; there was no way
to survey task runs. Adds run_list/run_events to the shared registry
(guarded as read on task kind), attaches the instance's acting principal
to curator model calls so guarded tools authorize, and records a pass's
final text as a pass.summary action so the report is inspectable. Ships a
dry-run self-intake definition in dev.toml/config.example.toml declaring
only those two tools, so it cannot file issues yet.
2026-11-03 02:35:06 +11:00
sam 443a7a501b refactor(storecontract): split the store contract into topical files (#1781)
Pure move: one 624-line storecontract.go becomes storecontract.go
(package doc + composed TaskStore), task.go, capture.go, binding.go,
mapping.go, source.go, status.go, resource.go, credentials.go and
errors.go. The package remains the single seam; no declaration changed
(verified line-identical against HEAD). task check green.
2026-11-03 02:27:57 +11:00
sam 88e86776e8 feat(forge): add CreateIssue to the forge contract (#1770)
Quality / check (push) Successful in 4m10s
deploy / gate (push) Successful in 4m43s
deploy / archied-image (push) Successful in 2m54s
deploy / agent-image (push) Successful in 11s
deploy / dist-zip (push) Skipped
Issue-level writes could comment and close but not open an issue, so no
curator or workflow stage can file its own work yet. Adds the rpc to
proto/forge/v1, regenerates the contracts, and threads it through the
domain interface, forgerpc client/server (comment permission, like
CloseIssue), forgeext, and the archipelago SDK + gitea/github plugins.
2026-11-03 02:04:56 +11:00
sam e10bb161fe docs(prds): mark the self-development ADR as decided, not built (#1769) 2026-11-03 01:52:06 +11:00
sam 3b525b7482 docs(prds): mark the self-development ADR as record, not mechanism (#1769) 2026-11-03 01:50:54 +11:00
sam c2cc10ebfa docs(prds): record the self-development loop decision (#1769) 2026-11-03 01:36:31 +11:00
sam 2f8076d7f8 fix(access): forbid clause scopes the canary grant against its developer role (#1727)
deploy / gate (push) Successful in 4m45s
Quality / check (push) Successful in 3m37s
deploy / archied-image (push) Successful in 2m53s
deploy / agent-image (push) Successful in 10s
deploy / dist-zip (push) Skipped
Service accounts created through the identities API are assigned the
developer role in the request's org, whose shipped policy permits read and
write across the org -- the permit alone did not scope the canary. Install a
matching org-canary-grant-forbid policy (access.CanaryGrantForbidPolicy) that
negates exactly the permit's surface for the canary identity; a forbid wins
within the level, so the role can no longer widen access. Engine tests pin
the surface with the principal carrying the developer role, and the
rehearsal now proves it live: the scoped token gets 403 on /api/access/policies
and 200 on /api/chat/update, and the full canary loop runs on it.

task check passes; task canary:rehearse green (REHEARSE_EXIT=0).
2026-11-02 06:23:38 +11:00
sam c180a2debb feat(access): scoped canary grant replaces instance-admin CANARY_TOKEN (#1727)
Quality / check (push) Successful in 3m49s
deploy / gate (push) Successful in 4m48s
deploy / archied-image (push) Successful in 2m55s
deploy / agent-image (push) Successful in 11s
deploy / dist-zip (push) Skipped
The canary now authenticates as a dedicated service-account identity in the
system org whose only grant is access.CalaryGrantPolicy: read the signal
endpoints (chat update, alerts, logs, tasks), read and write the
update-settings resource, and trigger an update install. CANARY_TOKEN no
longer carries instance admin; DEPLOY_TOKEN stays CI-only.

- authorize: POST /api/chat/update/install and /defer map to
  update on Object::"update" (dashboard) instead of run on any workflow,
  so the grant can cover installs without granting workflow runs
- accounts: active service-account identities may authenticate with bearer
  tokens (CanAct), not just users
- new POST /api/identities/{id}/tokens (behind manage_identities) mints a
  token for a named non-user identity; new AccountService RPC
  BindIdentityPersonalToken backs it remotely, refusing user-kind targets,
  with the caller recorded as audit actor
- engine tests pin the grant's exact allow/deny surface

RELEASING.md documents the one-time setup. task check passes on this tree.
2026-11-02 06:03:41 +11:00
sam 5356745968 fix(releaseupdate): address review findings on the #1725 canary change
Quality / check (push) Successful in 3m39s
deploy / gate (push) Successful in 4m28s
deploy / archied-image (push) Successful in 3m11s
deploy / agent-image (push) Successful in 11s
deploy / dist-zip (push) Skipped
- one shared UpdateStateDir() resolver for daemon and dashboard, instead of
  two copies that could drift
- the recovery path for a committed/rolled-back journal writes
  installed.json too, so a crash after the outcome phase cannot lose the
  install record
- JudgeCanaryNow is the explicit signals-only entry point for promoting
  without the window, replacing the zero-install-time sentinel
- canary version comparisons use IssuedAt, matching Promote's replay rule,
  so semver/issued-at divergence cannot wedge check red; queuedBehind
  reuses the judgeable predicate instead of a looser check
- promote prints the channel/pin; promote.yml cancels superseded runs
- canary card: no duplicated elapsed rendering, channel row waits for the
  settings resource instead of showing a default, agent fallback '?'
- rehearsal script bounds its docker build/push/run calls

task check passes on this tree.
2026-11-02 02:45:26 +11:00
sam 5b5059615b feat(releaseupdate): judge the installed release, window from install (#1725)
deploy / gate (push) Successful in 4m31s
deploy / archied-image (push) Successful in 3m14s
deploy / dist-zip (push) Skipped
Quality / check (push) Successful in 4m2s
deploy / agent-image (push) Successful in 11s
The canary window now counts from when a release is installed on the host
(installed.json / installed_at in the update snapshot), not from the
statement's issued_at, and 'release canary check' judges the release the
canary actually runs via its exact-pin tag instead of the moving next tag.
A newer release on next queues behind the one being judged instead of
restarting the clock, so stable moves once per release.

New subcommands: canary status (read-only state, signals, verdict), canary
promote (promote now when hard checks pass), canary unpin (back to next),
and start -force. A held/pinned canary is verified installable every check
and a stranded pin fails the run red instead of looking green. Rollback
pre-flight checks stable's zip and agent image digests before pinning.
Relative signals are judged whenever both windows carry minimum data, so a
breach rolls back mid-window; promote still needs the full window unless
the operator promotes early (new promote.yml workflow_dispatch). The window
is configurable via -window / CANARY_WINDOW.

Also: dashboard canary status card (ui), and task canary:rehearse, a local
end-to-end rehearsal of publish -> canary start -> status -> check ->
promote against a throwaway registry and host.

task check passes; rehearsal run green (REHEARSE_EXIT=0).
2026-11-02 01:15:29 +11:00
418 changed files with 9382 additions and 3924 deletions

No files matched your search

+12 -4
View File
@@ -1,7 +1,9 @@
# Every 15 minutes, checks how the newest release on the `next` channel is
# running on the canary host. If a hard check fails, the canary goes back to
# the stable release straight away. If the release stays healthy for 24 hours,
# it's promoted to stable. A rollback fails the run, so it shows up red.
# Every 15 minutes, checks how the release installed on the canary host is
# running, judged over the window counted from its install (-window, default
# 24h; CANARY_WINDOW overrides it). If a hard check fails, the canary goes
# back to the stable release straight away. If the release stays healthy for
# the window, it's promoted to stable. A rollback fails the run, so it shows
# up red.
name: canary
on:
@@ -48,6 +50,12 @@ jobs:
- name: Judge the canary
env:
ARCHIE_CANARY_TOKEN: ${{ secrets.CANARY_TOKEN }}
CANARY_WINDOW: ${{ vars.CANARY_WINDOW }}
run: |
window_args=()
if [ -n "$CANARY_WINDOW" ]; then
window_args=(-window "$CANARY_WINDOW")
fi
"$RUNNER_TEMP/archied" release canary check -dashboard "${{ vars.CANARY_URL }}" \
"${window_args[@]}" \
-latest git.catlow.cloud/sam/archied,git.catlow.cloud/sam/archie-agent
+6 -2
View File
@@ -418,8 +418,12 @@ jobs:
"$api/repos/$repo/releases" >/dev/null
echo "published release notes for $tag"
# The canary host follows the `next` channel. canary.yml then decides
# whether this release is promoted to stable or the canary is rolled back.
# The canary host follows the `next` channel. canary.yml then judges the
# release installed on the canary and decides whether it is promoted to
# stable or the canary is rolled back. The start is queue-safe: a release
# published while the canary still judges a newer one is queued behind
# it and picked up by the next check, so the running window isn't
# restarted under a different release.
- name: Install the release on the canary
if: vars.CANARY_URL != ''
env:
+55
View File
@@ -0,0 +1,55 @@
# Promotes the release the canary is running right now, without waiting out
# the judging window: the canary's signals must show no hard failure and no
# relative breach, otherwise the run refuses and fails. Dispatch it with the
# version the canary runs (`archied release canary status` shows it).
name: promote
on:
workflow_dispatch:
inputs:
version:
description: "release the canary is expected to run; must match what it runs"
required: true
type: string
concurrency:
group: promote
cancel-in-progress: true
permissions:
contents: read
packages: write
jobs:
promote:
if: vars.CANARY_URL != ''
runs-on: self-hosted
steps:
- name: Checkout
uses: actions/checkout@v4
with:
ref: main
- name: Set up Go
uses: actions/setup-go@v5
with:
go-version-file: go.mod
# The self-hosted runner keeps the Go caches on disk between jobs, and
# archiving them for the Actions cache takes minutes.
cache: false
- name: Build archied
# The promote never serves the dashboard, so the build leaves it out.
run: go build -tags no_ui -o "$RUNNER_TEMP/archied" ./cmd/archied
# Promotion retags stable and the images' latest in the registry.
- name: Log in to the Gitea registry
run: echo "${{ secrets.DEPLOY_TOKEN }}" | docker login git.catlow.cloud -u "${{ github.actor }}" --password-stdin
- name: Promote the canary's release
env:
ARCHIE_CANARY_TOKEN: ${{ secrets.CANARY_TOKEN }}
run: |
"$RUNNER_TEMP/archied" release canary promote -dashboard "${{ vars.CANARY_URL }}" \
-version "${{ inputs.version }}" \
-latest git.catlow.cloud/sam/archied,git.catlow.cloud/sam/archie-agent
+5 -9
View File
@@ -20,12 +20,11 @@ jobs:
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
# ripgrep and the unprivileged-userns sysctl are provisioned on the host.
# A self-hosted runner should not need sudo, and installing a package on
# every run is not the gate's job.
- name: Verify unprivileged user namespaces
run: unshare --user --map-root-user --net true
- uses: actions/setup-go@v5
with:
@@ -46,9 +45,6 @@ jobs:
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
- name: Ensure contract baseline ref
run: |
if ! git show-ref --verify --quiet refs/heads/main; then
+11
View File
@@ -21,6 +21,17 @@ jobs:
cut:
runs-on: self-hosted
steps:
# A mirror carries its own Actions secrets. An empty token checks out
# fine and fails at the push instead, which reads as a broken release.
- name: Check DEPLOY_TOKEN is configured
env:
DEPLOY_TOKEN: ${{ secrets.DEPLOY_TOKEN }}
run: |
if [ -z "$DEPLOY_TOKEN" ]; then
echo "::error::DEPLOY_TOKEN is not set for this repository, so the release cannot push its commit and tag. Set the secret here, or dispatch the release from the Gitea primary."
exit 1
fi
# DEPLOY_TOKEN rather than the job token: a push made with the job token
# doesn't start workflows, and the tag push must start deploy.yml.
- name: Checkout
+1
View File
@@ -47,3 +47,4 @@ __pycache__/
# Local symlink to the archipelago marketplace repo
/archipelago
.agents-run/
+28
View File
@@ -5,6 +5,34 @@ release's per-component sections are labelled beneath its heading.
## [Unreleased]
## [1.61.0] - 2026-11-07
### archied
- A `workflow.handoff` now finishes its run and starts a separate, linked run of the next workflow on the same issue, instead of continuing as another attempt. The task page links each run to the one before and after it; attempts count only retries of one run.
- A task no longer fails to prepare its worktree with "some refs were not updated" after a tag or branch moves on the forge.
### archie-ui
- The task page no longer flashes its loading state on every live update while a task runs.
### archie-state-store
- The extension catalogue lists and installs again. Its default signing key has changed; an instance that set its own catalogue signing key in Extension settings keeps using that key.
## [1.60.0] - 2026-11-02
### archied
- Curators can list a repository's open issues and file new ones with the `issue_list` and `issue_create` tools, which reach the forge through the daemon. A curator files only when its definition declares `issue_create`; the shipped self-intake example runs in dry run with `issue_list` for deduplication. Listing every open issue needs the forge extensions from the archipelago release after 1.2.0.
## [1.59.0] - 2026-11-02
### archied
- Forge extensions can open issues: the forge surface gains `CreateIssue` alongside list, comment and close, and the bundled Gitea and GitHub extensions implement it.
## [1.58.1] - 2026-11-01
### archied
+137 -20
View File
@@ -47,6 +47,21 @@ then pushes `main` and the tag atomically with `DEPLOY_TOKEN`. A push made with
the job token wouldn't start other workflows, so the user token is what lets
the tag start the deploy.
Every workflow that reaches the registry or the canary reads its credentials
from Actions secrets, so a mirror of this repository carries its own set:
| Name | Kind | Used by |
|---|---|---|
| `DEPLOY_TOKEN` | secret | release, deploy, canary, promote — the registry push, the tag push, and the package links |
| `RELEASE_SIGNING_KEY` | secret | deploy — signs the release statement |
| `CANARY_TOKEN` | secret | deploy, canary, promote — the canary's dashboard API |
| `CANARY_URL` | variable | deploy, canary, promote — where that dashboard is |
| `CANARY_WINDOW` | variable | canary — the judged window; empty means the default |
A mirror missing them fails at its first login or push; `release.yml` names the
missing `DEPLOY_TOKEN` before it checks out, so the run doesn't read as a broken
release.
After a release, pull `main` before committing on it again: CI has added the
release commit.
@@ -78,27 +93,64 @@ release rather than committing the fix on top of a tagged one.
The canary is the production host, following `next`. Every quarter hour
`canary.yml` runs `archied release canary check`, which reads the canary
through its dashboard API and judges the release on `next`:
through its dashboard API and judges **the release the canary actually runs**,
not the moving `next` tag:
| Signal | Source | Fails when |
| --- | --- | --- |
| Install outcome | `GET /api/chat/update` | the canary doesn't run the release (the watchdog rolled it back) |
| Install outcome | `GET /api/chat/update` | the canary doesn't run the judged release (the watchdog rolled it back) |
| Readiness | `GET /health/detailed` | the rollup isn't `ok` |
| Alerts | `GET /api/alerts` | an open `action` alert was first raised after the release was issued |
| Error log | `GET /api/logs?level=error` | error lines per day exceed `2 × baseline + 10` |
| Alerts | `GET /api/alerts` | an open `action` alert was first raised after the release was installed |
| Error log | `GET /api/logs?level=error` | error lines per window exceed `2 × baseline + 10` |
| Task outcomes | `GET /api/tasks` | the share of finished tasks that parked or died rose more than 10 points |
- **Control.** The canary's own 24 hours before the release was issued. With
- **Control.** The canary's own window before the release was installed. With
one production host there's no simultaneous control.
- **Window.** 24 hours from issue, one cycle of schedules and traffic. The
first 30 minutes are for the install and restart.
- **Window.** 24 hours counted from the **install** on the canary, not from
the release's issue time: the host records it in `installed.json`
(`committed_at`) and reports it as `installed_at` in `GET /api/chat/update`.
One cycle of schedules and traffic. The first 30 minutes are for the install
and restart. `canary.yml` passes `-window` when the `CANARY_WINDOW` Actions
variable is set; the flag defaults to 24h. A host that doesn't report an
install time degrades to the release statement's issue time.
- **Hard signals** (install, readiness, alerts) roll back at any check after
that. **Relative signals** (errors, tasks) are judged at the end of the
window.
that. **Relative signals** (errors, tasks) are judged against the baseline
whenever both windows carry enough data, so a breach rolls the release back
even mid-window; a full window with no failure promotes it.
- **Too little data skips a signal** rather than failing it: fewer than 5
finished tasks in the window, or an error log that reaches back less than an
hour into the baseline. A log read that stops short of a window is scaled to
a full day from what it covers.
a full window from what it covers.
### Queueing
`deploy.yml` installs each new release on the canary with
`archied release canary start -version V`, which first points the canary back
at `next` (clearing any rollback pin) and then installs `V` — unless the
canary is already judging a release newer than stable that isn't `V`, in which
case `V` is **queued** behind it and the canary keeps running what it judges.
`-force` installs regardless. A `check` that promotes (or finds nothing to
judge) installs the newer `next` in the same run, so a queue drains one
release per promotion.
### Subcommands
```
archied release canary status read-only: channel/pin, installed release and
installed_at, stable/next, window progress, and
the verdict the current signals would give
archied release canary start point the canary at next and install -version,
queued behind a release being judged (-force to
override)
archied release canary check judge the installed release over -window
(default 24h): promote, pick up a newer next,
or roll back; a held canary is reported, and a
held pin that is no longer installable fails
the run red
archied release canary promote promote the installed release now if its
signals pass; refuses with exit 1 otherwise
archied release canary unpin return the canary to next from any pin
```
### Promote
@@ -106,32 +158,97 @@ A release that passes the window is promoted: `stable` is retagged to the
`v<version>` manifest. The signature covers the statement, not the tag, so the
retag needs no signing key. The `archied` and `archie-agent` images' `:latest`
move to their `:v<version>` first, and the agent image must be the digest the
statement pins. Promotion refuses a prerelease and a release issued
before the one `stable` names, the same replay rule hosts enforce. The decision
and its reasons are in the `canary` run log.
statement pins. Promotion refuses a prerelease and a release issued before the
one `stable` names, the same replay rule hosts enforce. The decision and its
reasons are in the `canary` run log.
Early promotion is a dispatch away: the **promote** workflow
(`.github/workflows/promote.yml`) takes a `version` input and runs
`archied release canary promote -version V` on the canary. It judges the
release the canary runs — the window doesn't matter for the decision — and
refuses (red run) on any hard failure or relative breach with enough data.
### Roll back
`stable` stays put, and the run turns red. The canary is pinned to the `stable`
version through its update settings and reinstalled through its dashboard.
`stable` stays put, and the run turns red. Before the canary is pinned to
`stable`, `check` verifies `stable` is actually installable — its statement's
zip blob must still be in the release repository and its pinned agent image
must still resolve (`releaseupdate.Installable`). If it isn't, the run fails
loudly and the canary stays on `next` rather than being stranded on a pin.
Otherwise the canary is pinned to the `stable` version through its update
settings and reinstalled through its dashboard.
A release whose migrations differ from `stable`'s can't go back without
restoring the pre-update database snapshot, which loses every write since. The
canary is held on the failed release instead (pinned to it) and the fix goes
forward in the next release. The next release's deploy points the canary back at
`next`.
forward in the next release. The next release's deploy points the canary back
at `next`.
A pinned canary isn't judged, so a rollback is decided once.
A pinned canary isn't judged, so a rollback is decided once — but `check`
verifies the pin is still installable every run, so a stranded pin shows red
instead of green. `archied release canary unpin` returns the canary to `next`
by hand.
### Configuration
| Name | Kind | Holds |
| --- | --- | --- |
| `CANARY_URL` | Actions variable | the canary's dashboard URL; unset disables the canary steps |
| `CANARY_TOKEN` | Actions secret | a personal API token on the canary with instance admin |
| `DEPLOY_TOKEN` | Actions secret | registry and repository write, for publishing, promoting and the release push |
| `CANARY_TOKEN` | Actions secret | an API token bound to a dedicated `service_account` identity on the canary, carrying only the canary grant policy (see below) — never instance admin |
| `CANARY_WINDOW` | Actions variable | optional `-window` override for `canary check`; unset is 24h |
| `DEPLOY_TOKEN` | Actions secret | registry and repository write, for publishing, promoting and the release push; stays CI-only, promotion runs in CI and never on the host |
| `RELEASE_SIGNING_KEY` | Actions secret | the release signing seed |
#### Setting up `CANARY_TOKEN` once
`CANARY_TOKEN` is a bearer token for a dedicated service-account identity in
the system org whose only grant is the canary policy
(`access.CanaryGrantPolicy`): read the signal endpoints (`GET /api/chat/update`,
`GET /api/alerts`, `GET /api/logs`, `GET /api/tasks`), read and write the
`update-settings` resource, and trigger an update install
(`POST /api/chat/update/install`). It carries no other rights, and instance
admin is never needed. With an instance-admin token (for example the operator
session), run against the canary host:
```bash
# 1. Create the service-account identity in the system org.
curl -X POST "$CANARY_URL/api/identities" -H "Authorization: Bearer $ADMIN" \
-H 'Content-Type: application/json' \
-d '{"kind":"service_account","display_name":"ci-canary"}'
# -> {"id":"<canary-identity-id>", ...}
# 2. Issue its token (shown exactly once).
curl -X POST "$CANARY_URL/api/identities/<canary-identity-id>/tokens" \
-H "Authorization: Bearer $ADMIN" -H 'Content-Type: application/json' \
-d '{"name":"ci-canary","days":90}'
# -> {"id":"<token-hash>","token":"archie_pat_..."} -- store the token value as the CANARY_TOKEN secret
# 3. Install the grant pair, scoped to that identity. The policies' org is
# pinned to the caller's org (org-sys for the instance admin), so the
# bodies need no org field. The permit alone is not enough: service
# accounts created through the API are assigned the developer role, so a
# matching forbid (installed second) negates everything outside the
# permit's surface for that identity.
curl -X PUT "$CANARY_URL/api/access/policies" -H "Authorization: Bearer $ADMIN" \
-H 'Content-Type: application/json' \
-d '{"id":"org-canary-grant","level":"org","text":"permit(principal, action, resource) when {\n resource.org == principal.org &&\n principal == Archie::Identity::\"<canary-identity-id>\" &&\n (\n (action == Archie::Action::\"read\" && [\"dashboard\", \"task\"].contains(resource.kind)) ||\n (action == Archie::Action::\"read\" && resource == Archie::Object::\"update-settings\") ||\n (action == Archie::Action::\"update\" && resource == Archie::Object::\"update\") ||\n (action == Archie::Action::\"manage_identities\" && resource == Archie::Object::\"update-settings\")\n )\n};"}'
```
The second policy forbids the canary identity from everything the permit does
not name, so its developer role cannot widen the grant:
```bash
curl -X PUT "$CANARY_URL/api/access/policies" -H "Authorization: Bearer $ADMIN" \
-H 'Content-Type: application/json' \
-d '{"id":"org-canary-grant-forbid","level":"org","text":"forbid(principal, action, resource) when {\\n resource.org == principal.org &&\\n principal == Archie::Identity::\\\"<canary-identity-id>\\\" &&\\n !(\\n (action == Archie::Action::\\\"read\\\" && [\\\"dashboard\\\", \\\"task\\\"].contains(resource.kind)) ||\\n (action == Archie::Action::\\\"read\\\" && resource == Archie::Object::\\\"update-settings\\\") ||\\n (action == Archie::Action::\\\"update\\\" && resource == Archie::Object::\\\"update\\\") ||\\n (action == Archie::Action::\\\"manage_identities\\\" && resource == Archie::Object::\\\"update-settings\\\")\\n )\\n};"}'
```
The policy texts are exactly `access.CanaryGrantPolicy` and
`access.CanaryGrantForbidPolicy` from `internal/domain/access/canary.go` with
the identity id filled in. The canary needs no other rights: `GET
/health/detailed` is unauthenticated, and every other call it makes is covered
by the grant pair.
The runner must reach `CANARY_URL`. Without a canary a release stays on `next`
until promoted by hand, under the same refusals:
+7
View File
@@ -221,6 +221,13 @@ tasks:
cmds:
- node e2e/run.cjs {{.CLI_ARGS}}
canary:rehearse:
# Local rehearsal only: spins up its own registry, database and host, and
# is deliberately absent from check and CI.
desc: Rehearse the release-canary loop on a throwaway local host (needs docker; never part of the gate)
cmds:
- sh tools/canary-rehearse.sh
build:
desc: Build archie services and agent binaries
deps: [ui]
+9
View File
@@ -156,6 +156,15 @@ _needs a repository._
| `push` | bool | Push the branch after committing. |
| `reference` | `Fixes` \| `Implements` \| `Refs` | How the commit links to the task issue. |
Hooks never run. The commit is created by go-git, which ignores `.git/hooks`
and `core.hooksPath`, and the agent never runs git, so `pre-commit`,
`commit-msg` and `prepare-commit-msg` in the checkout are inert files. A
repository whose rules live only in a hook — a formatter, a secret scanner,
commitlint, a DCO trailer — gets a commit that does not satisfy them: the run
reports success and the pull request fails the check afterwards. Put the same
commands in `command.run` or the repository gate, which both run in the task
container before this step.
### `repo.open-pr`
_needs a repository._
+17
View File
@@ -94,3 +94,20 @@ target = "127.0.0.1:9090"
# releases_key = "base64 public key"
# catalogue = "registry.example/owner/archipelago/catalogue"
# catalogue_key = "base64 public key"
# Long-running agent loops that maintain memory, skills and — in dry run —
# file follow-up issues from recent runs. Definitions are data: the server
# validates them and the gateway applies them live.
# [[curators]]
# name = "self-intake"
# interval = "1h"
# tools = ["run_list", "run_events", "issue_list"]
# instructions = """
# You are archie's self-intake curator, in dry run. Each pass: list recent
# task runs with run_list, inspect anything parked or failed with run_events,
# check issue_list for each repo so nothing already open is repeated, then
# report candidate follow-up issues, each tied to the run that produced it.
# Do not create issues.
# """
# To let it file: add "issue_create" to tools and tell it to file each
# candidate (citing the run) instead of only reporting it.
+11
View File
@@ -45,3 +45,14 @@ listen = "127.0.0.1:8585"
[services.daemon]
target = "127.0.0.1:8587"
listen = "127.0.0.1:8587"
# Dry-run self-intake curator (#1770): surveys recent runs and reports
# candidate follow-up issues as pass summaries; declares no issue-creating
# tool, so it cannot file. Needs a chat model configured to run at all.
[[curators]]
name = "self-intake"
interval = "1h"
tools = ["run_list", "run_events", "issue_list"]
instructions = """
You are archie's self-intake curator, in dry run. Each pass: list recent task runs with run_list, inspect anything parked or failed with run_events, check issue_list for each repo so nothing already open is repeated, then report candidate follow-up issues — concrete, each tied to the run that produced it. Do not create issues; this pass reports only (no issue-creating tool is declared, so it cannot). End your reply with the report.
"""
+6
View File
@@ -88,7 +88,10 @@
"/news/archie/1.57.0/": "/news/1.57.0/",
"/news/archie/1.58.0/": "/news/1.58.0/",
"/news/archie/1.58.1/": "/news/1.58.1/",
"/news/archie/1.59.0/": "/news/1.59.0/",
"/news/archie/1.6.0/": "/news/1.6.0/",
"/news/archie/1.60.0/": "/news/1.60.0/",
"/news/archie/1.61.0/": "/news/1.61.0/",
"/news/archie/1.7.0/": "/news/1.7.0/",
"/news/archie/1.8.0/": "/news/1.8.0/",
"/news/archie/1.9.0/": "/news/1.9.0/",
@@ -192,7 +195,10 @@
"/news/archied/1.57.0/": "/news/1.57.0/",
"/news/archied/1.58.0/": "/news/1.58.0/",
"/news/archied/1.58.1/": "/news/1.58.1/",
"/news/archied/1.59.0/": "/news/1.59.0/",
"/news/archied/1.6.0/": "/news/1.6.0/",
"/news/archied/1.60.0/": "/news/1.60.0/",
"/news/archied/1.61.0/": "/news/1.61.0/",
"/news/archied/1.7.0/": "/news/1.7.0/",
"/news/archied/1.8.0/": "/news/1.8.0/",
"/news/archied/1.9.0/": "/news/1.9.0/",
+15
View File
@@ -1,4 +1,19 @@
[
{
"version": "1.61.0",
"date": "2026-11-07",
"body": "### archied\n\n- A `workflow.handoff` now finishes its run and starts a separate, linked run of the next workflow on the same issue, instead of continuing as another attempt. The task page links each run to the one before and after it; attempts count only retries of one run.\n\n- A task no longer fails to prepare its worktree with \"some refs were not updated\" after a tag or branch moves on the forge.\n\n### archie-ui\n\n- The task page no longer flashes its loading state on every live update while a task runs.\n\n### archie-state-store\n\n- The extension catalogue lists and installs again. Its default signing key has changed; an instance that set its own catalogue signing key in Extension settings keeps using that key."
},
{
"version": "1.60.0",
"date": "2026-11-02",
"body": "### archied\n\n- Curators can list a repository's open issues and file new ones with the `issue_list` and `issue_create` tools, which reach the forge through the daemon. A curator files only when its definition declares `issue_create`; the shipped self-intake example runs in dry run with `issue_list` for deduplication. Listing every open issue needs the forge extensions from the archipelago release after 1.2.0."
},
{
"version": "1.59.0",
"date": "2026-11-02",
"body": "### archied\n\n- Forge extensions can open issues: the forge surface gains `CreateIssue` alongside list, comment and close, and the bundled Gitea and GitHub extensions implement it."
},
{
"version": "1.58.1",
"date": "2026-11-01",
+68
View File
@@ -0,0 +1,68 @@
# Self-development loop: archie develops its own codebase
**Status:** Decided, not yet built (2026-11-03, sam) **Scope:** how archie
persistently takes on development of archie-core itself. This file records a
desired state, not an implementation or a product: nothing below exists unless
the "Built today" section says so. The plan is tracked as epic #1769 with child
issues #1770–#1775. This file is a record, not a mechanism: agents only see
what is in their context, so verify every claim here against the code before
acting on it — where they disagree, the code wins. The boundary below is
load-bearing only where it is enforced (the gate, access control, the daemon's
git monopoly) and stated in `AGENTS.md`; everywhere else this file is a pointer.
## Built today
The machinery the loop will reuse exists: shipped triage/implement/pr-review
workflows, the adversarial self-review stage before a PR opens (epic #1348,
closed), the prbench judge for review quality with its acceptance bar, canary
releases judged on the installed release (releaseupdate), and the curator
engine. What does not exist is the loop itself: no self-intake curator, no
approved-issue auto-run, no judge for diffs, no outcome distillation, no
canary routing specific to self-directed changes.
## Decision
Archie improves its own codebase through a closed loop, built in stages:
1. **Self-intake** (#1770): a curator pass reads run outcomes, step transcripts,
gate failures and alerts, dedupes against open issues, and files new issues.
Archie files its own work; the maintainer approves priority.
2. **Approved-issue loop** (#1771): an approved issue runs the shipped
triage/implement workflows end to end — read-only plan, gated build,
adversarial self-review, daemon commits and opens a green PR. The maintainer
still merges at this stage.
3. **Implementation judge** (#1772): the prbench judge pattern scores diffs,
calibrated against the maintainer's merge decisions; auto-merge applies only
when the gate is green, the review stage is clean, and the judge is above the
calibrated bar.
4. **Outcome distillation** (#1773): a curator turns merged-run postmortems into
proposals for the repo's agent instructions, each added only because it
changes a decision; rules that stop changing decisions are removed.
5. **Canary** (#1774): self-directed changes ship through the canary release
window and are judged on the installed release before promotion. Rollback is
a plain revert, because every self-directed change is one gated commit.
## Boundary
Enforcement lives in the harness, never in the prompt (#1775). Changes to the
commit gate, the invariants in `AGENTS.md`, access control and grant scoping,
the daemon-only git path, and webhook HMAC verification may be proposed by
archie but are never merged by archie — they require a maintainer merge.
Everything else (workflows, curators, memory, docs, the dashboard) is fair game
for the loop.
## Why
Self-improvement degrades when it relies on the model choosing to follow rules.
Gates, scoped grants and the daemon's monopoly on git cannot be argued with; a
prompt can. Keeping quality measurable — a judge bar and a canary verdict as
numbers, tests guarding the protected seams — is what lets the loop widen
without turning the codebase to garbage.
## Consequences
- One issue → one small PR applies doubly to self-directed work: past about 100
lines or any new mechanism, the change stops and reassesses.
- Every self-directed commit cites the issue that authorized it.
- Rule proposals from the distillation curator must name the run that triggered
them, and `AGENTS.md` stays short enough to read in full every session.
+126 -3
View File
@@ -1,9 +1,14 @@
// Keyboard and phone access: the task dialog fits and tabs in order, pages
// title correctly, the skip link and palette focus main content, and
// query-only tab switches keep focus where it is.
// query-only tab switches keep focus where it is. Then the audit's two
// static sweeps (#1767): no unnamed controls on any route, and every keyboard
// focus stop's indicator clears 3:1 against the background behind it.
const assert = require('node:assert/strict');
const { staticRoutes } = require('./helpers/routes.cjs');
module.exports = async ({ page, go, shot, settled }) => {
module.exports = async ({ page, go, shot, fail, settled, tolerate }) => {
// The extension catalogue 502 is a stack-environment issue, not this check.
tolerate(/extensions\/catalogue/);
await page.setViewportSize({ width: 390, height: 844 });
await go('/tasks');
await page.getByRole('button', { name: 'Jump to', exact: true }).click();
@@ -35,7 +40,7 @@ module.exports = async ({ page, go, shot, settled }) => {
assert.ok(await skip.evaluate(el => el === document.activeElement), 'Skip link is not the first keyboard stop');
await page.keyboard.press('Enter');
assert.equal(await page.evaluate(() => document.activeElement?.id), 'main-content');
await page.getByRole('button', { name: 'Sections', exact: true }).click();
await page.getByRole('button', { name: 'Toggle Sidebar', exact: true }).click();
await page.getByRole('navigation', { name: 'Sections', exact: true }).getByRole('link', { name: 'Tasks', exact: true }).click();
await page.waitForFunction(() => document.activeElement?.id === 'main-content', null, { timeout: 5000 }).catch(() => {});
assert.equal(await page.evaluate(() => document.activeElement?.id), 'main-content', 'route navigation did not focus main');
@@ -50,4 +55,122 @@ module.exports = async ({ page, go, shot, settled }) => {
const tab = page.getByRole('tab', { name: /^Map/ });
await tab.click();
assert.ok(await tab.evaluate(el => el === document.activeElement), 'query-only navigation stole focus');
await page.setViewportSize({ width: 1440, height: 900 });
// Unnamed controls: every visible input, textarea, select, button or
// widget role on each static route needs an accessible name (accname
// order: labelledby, label, content, value, placeholder, title).
for (const r of staticRoutes()) {
await go(r);
const unnamed = await page.evaluate(() => {
const name = (el) => {
const by = el.getAttribute('aria-labelledby');
if (by) {
const t = by.split(/\s+/).map((id) => (document.getElementById(id)?.textContent || '')).join(' ').trim();
if (t) return t;
}
const al = el.getAttribute('aria-label');
if (al && al.trim()) return al.trim();
if (el.id) {
const l = document.querySelector(`label[for="${CSS.escape(el.id)}"]`);
if (l && l.textContent.trim()) return l.textContent.trim();
}
const wrap = el.closest('label');
if (wrap && wrap.textContent.trim()) return wrap.textContent.trim();
if (el.tagName === 'BUTTON' || el.getAttribute('role') === 'button' || el.getAttribute('role') === 'tab' || el.getAttribute('role') === 'switch') return el.textContent.trim();
if (el.tagName === 'INPUT') {
const t = (el.getAttribute('type') || 'text').toLowerCase();
if (t === 'submit' || t === 'button' || t === 'reset') return el.value.trim();
if (t === 'image') return (el.getAttribute('alt') || '').trim();
}
return (el.getAttribute('placeholder') || el.getAttribute('title') || '').trim();
};
return [...document.querySelectorAll('button, input, textarea, select, [role="button"], [role="tab"], [role="switch"], [role="checkbox"]')]
.filter((el) => el.type !== 'hidden' && el.getClientRects().length)
.filter((el) => !name(el))
.map((el) => `${el.tagName.toLowerCase()}${el.id ? '#' + el.id : ''}: ${el.outerHTML.slice(0, 120)}`);
});
for (const u of unnamed) fail(`unnamed control on ${r}: ${u}`);
}
// Focus contrast: Tab through each static route; the focused element's
// outline or ring must clear 3:1 against the effective background behind it.
const focusProbe = () => page.evaluate(() => {
const el = document.activeElement;
if (!el || el === document.body || el === document.documentElement) return null;
const cs = getComputedStyle(el);
const parse = (c) => {
c = String(c).trim();
let m = c.match(/rgba?\(([^)]+)\)/);
if (m) {
const p = m[1].split(',').map(Number);
return { r: p[0], g: p[1], b: p[2], a: p.length > 3 ? p[3] : 1 };
}
m = c.match(/^#([0-9a-f]{3,8})$/i);
if (m) {
let h = m[1];
if (h.length === 3 || h.length === 4) h = [...h].map((x) => x + x).join('');
const n = parseInt(h.slice(0, 6), 16);
return { r: (n >> 16) & 255, g: (n >> 8) & 255, b: n & 255, a: h.length === 8 ? parseInt(h.slice(6, 8), 16) / 255 : 1 };
}
return null;
};
const bgBehind = (start) => {
for (let n = start; n; n = n.parentElement) {
const c = parse(getComputedStyle(n).backgroundColor);
if (c && c.a > 0) return c;
}
return { r: 255, g: 255, b: 255, a: 1 };
};
// An outer ring or outline is painted over the parent's background; an
// inset one over the element's own. Score against whichever is higher.
const bgSelf = bgBehind(el);
const bgParent = bgBehind(el.parentElement);
const over = (fg, bg) => {
const a = fg.a + bg.a * (1 - fg.a);
if (a === 0) return { r: 0, g: 0, b: 0 };
return { r: (fg.r * fg.a + bg.r * bg.a * (1 - fg.a)) / a, g: (fg.g * fg.a + bg.g * bg.a * (1 - fg.a)) / a, b: (fg.b * fg.a + bg.b * bg.a * (1 - fg.a)) / a };
};
const lum = (c) => {
const f = (v) => { v /= 255; return v <= 0.03928 ? v / 12.92 : Math.pow((v + 0.055) / 1.055, 2.4); };
return 0.2126 * f(c.r) + 0.7152 * f(c.g) + 0.0722 * f(c.b);
};
const ratio = (c1, c2) => {
const l1 = lum(c1), l2 = lum(c2);
return (Math.max(l1, l2) + 0.05) / (Math.min(l1, l2) + 0.05);
};
let fg = null;
// The ring lives in --tw-ring-shadow on :focus-visible; read the var
// directly, then fall back to outline, any shadow layer, focus border.
if (el.matches(':focus-visible')) {
const m = cs.getPropertyValue('--tw-ring-shadow').match(/rgba?\([^)]+\)|#[0-9a-f]{3,8}\b/i);
if (m) { const c = parse(m[0]); if (c && c.a > 0) fg = c; }
}
if (!fg && cs.outlineStyle !== 'none' && parseFloat(cs.outlineWidth) > 0) fg = parse(cs.outlineColor === 'currentcolor' ? cs.color : cs.outlineColor);
if (!fg && cs.boxShadow && cs.boxShadow !== 'none') {
for (const m of cs.boxShadow.matchAll(/rgba?\([^)]+\)|#[0-9a-f]{3,8}\b/gi)) {
const c = parse(m[0]);
if (c && c.a > 0) { fg = c; break; }
}
}
if (!fg && el.matches(':focus-visible') && parseFloat(cs.borderTopWidth) > 0) {
const c = parse(cs.borderTopColor);
if (c && c.a > 0) fg = c;
}
const sig = `${el.tagName}#${el.id}:${(el.textContent || '').trim().slice(0, 30)}`;
if (!fg) return { sig, ratio: 0 };
return { sig, ratio: Math.max(ratio(over(fg, bgSelf), bgSelf), ratio(over(fg, bgParent), bgParent)) };
});
for (const r of staticRoutes()) {
await go(r);
let prev = '';
for (let i = 0; i < 40; i++) {
await page.keyboard.press('Tab');
const info = await focusProbe();
if (!info || info.sig === prev) break;
prev = info.sig;
if (info.ratio < 3) fail(`focus indicator under 3:1 on ${r} at ${info.sig} (ratio ${info.ratio.toFixed(2)})`);
}
}
};
+19
View File
@@ -0,0 +1,19 @@
// The dashboard's static routes, read from the router source so a new page is
// swept automatically. Parameterised and redirect routes are skipped.
const fs = require('fs');
const path = require('path');
function staticRoutes() {
const src = fs.readFileSync(path.join(__dirname, '../../../ui/src/router/index.ts'), 'utf8');
const out = [];
for (const line of src.split('\n')) {
if (/redirect:/.test(line)) continue;
const m = line.match(/path: "([^"]*)"/);
if (!m || m[1].includes(':')) continue;
if (m[1].startsWith('/')) out.push(m[1]);
else if (m[1]) out.push('/org/' + m[1]);
}
return [...new Set(out)];
}
module.exports = { staticRoutes };
+18 -1
View File
@@ -1,5 +1,7 @@
const assert=require('node:assert/strict');
module.exports=async({page,go,shot})=>{
module.exports=async({page,go,shot,tolerate})=>{
// The extension catalogue 502 is a stack-environment issue, not this check.
tolerate(/extensions\/catalogue/);
await page.route('**/api/access/policies',async route=>{await new Promise(r=>setTimeout(r,2200));await route.continue()});
await go('/access/policies');
assert.equal(await page.getByText('No policies yet.',{exact:true}).count(),0);
@@ -16,4 +18,19 @@ module.exports=async({page,go,shot})=>{
await page.getByRole('button',{name:'Retry runs',exact:true}).click();
await page.getByRole('alert').filter({hasText:'Could not load recent runs'}).waitFor({state:'hidden'});
await shot('recovered');
// A failed list read is an error, never the empty state (#1767): force each
// page's list API to 500 and check its empty-state copy stays hidden.
const failList = async (pattern,path,emptyTexts,errorSel='[role=alert]')=>{
await page.route(pattern,route=>route.fulfill({status:200,contentType:'application/json',body:'unreadable response'}));
await go(path);
await page.locator(errorSel).first().waitFor();
for(const text of emptyTexts) assert.equal(await page.getByText(text,{exact:true}).count(),0,`${path} shows its empty state ("${text}") while the list read fails`);
await page.unroute(pattern);
};
await failList('**/api/tasks*','/tasks',['No tasks yet'],'text=Archie is not responding');
await failList('**/api/captures*','/events',['No captures yet']);
await failList('**/api/event-types*','/events',['No event types yet']);
await failList('**/api/tokens*','/account',['No tokens yet']);
await failList('**/api/access/policies*','/org/policies',['No policies yet']);
await shot('failed-reads-show-errors');
};
+6 -5
View File
@@ -1,4 +1,4 @@
// The topbar org selector: choosing a member org reloads the dashboard into it,
// The sidebar org switcher: choosing a member org reloads the dashboard into it,
// Tasks and Workflows then show only that org's work, and the API refuses an
// org the caller is not a member of.
const assert = require('node:assert/strict');
@@ -8,11 +8,12 @@ module.exports = async ({ page, go, shot, settled, tolerate }) => {
tolerate(ABSENT);
await go('/tasks');
await page.getByText('Fix flaky login test').first().waitFor();
const picker = page.getByRole('combobox', { name: 'Organisation' });
const picker = page.getByRole('button', { name: /^Organisation: / });
await picker.waitFor();
const offered = await (async () => {
await picker.click();
const names = await page.getByRole('option').allInnerTexts();
await page.getByRole('menuitem').first().waitFor();
const names = (await page.getByRole('menuitem').allInnerTexts()).map(t => t.trim().split('\n').pop().trim());
await page.keyboard.press('Escape');
return names;
})();
@@ -22,9 +23,9 @@ module.exports = async ({ page, go, shot, settled, tolerate }) => {
await shot('home-org-workflows');
await picker.click();
await Promise.all([page.waitForEvent('load'), page.getByRole('option', { name: 'Beta', exact: true }).click()]);
await Promise.all([page.waitForEvent('load'), page.getByRole('menuitem', { name: 'Beta' }).click()]);
await settled('/workflows');
await page.getByRole('combobox', { name: 'Organisation' }).waitFor();
await page.getByRole('button', { name: /^Organisation: / }).waitFor();
assert.equal(await page.getByText('Parked', { exact: true }).count(), 0, 'workflow run counts leaked across orgs');
await shot('beta-workflows');
await go('/tasks');
+69
View File
@@ -0,0 +1,69 @@
// Phone-width sweep of the views the November audit flagged (#1757): at
// 390x844 each must fit without horizontal scroll or clipped controls. The
// topbar (with its instance badge) is on every page; /account holds the
// token form, /org/agents the agent rows and the grants card, /events the
// sources table, the event-type rule editor and the mapping editor dialog.
const assert = require('node:assert/strict');
const { staticRoutes } = require('./helpers/routes.cjs');
module.exports = async ({ page, go, shot, fail, settled, tolerate }) => {
// The extension catalogue 502 is a stack-environment issue, not this check.
tolerate(/extensions\/catalogue/);
await page.setViewportSize({ width: 390, height: 844 });
const assertNoOverflow = async (label) => {
const { scrollWidth, innerWidth } = await page.evaluate(() => ({
scrollWidth: document.documentElement.scrollWidth,
innerWidth: window.innerWidth,
}));
if (scrollWidth > innerWidth) {
fail(`${label} overflows 390px: scrollWidth=${scrollWidth} > innerWidth=${innerWidth}`);
}
};
const visit = async (path, label, waitFor) => {
await go(path);
if (waitFor) await settled(waitFor);
await assertNoOverflow(label);
await shot(label);
};
await visit('/account', 'account-tokens', 'text=API tokens');
await visit('/org/agents', 'org-agents', 'text=Credential grants');
await go('/events');
await settled('[role=tab]');
await assertNoOverflow('events');
await shot('events');
// The seeded event type's rule editor (RuleEditor's condition row) opens
// from the Edit button on its row.
const editType = page.locator('button[aria-label^="Edit "]').first();
if (await editType.count()) {
await editType.click();
await page.getByRole('dialog').first().waitFor();
await settled();
await assertNoOverflow('event-type rule editor');
await shot('rule-editor');
await page.keyboard.press('Escape');
await settled();
}
// The mappings tab's editor dialog carries the payload tree.
await page.getByRole('tab', { name: 'Map' }).click();
await settled('text=New mapping');
await assertNoOverflow('mappings list');
const editMapping = page.locator('button', { hasText: 'Edit' }).first();
if (await editMapping.count()) {
await editMapping.click();
await page.getByRole('dialog').first().waitFor();
await settled();
await assertNoOverflow('mapping editor dialog');
await shot('mapping-editor');
}
// Every static route fits 390px (#1767): horizontal scroll fails.
for (const r of staticRoutes()) {
await visit(r, `route ${r}`, null);
}
};
+2 -16
View File
@@ -1,24 +1,10 @@
// 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)];
}
const { staticRoutes } = require('./helpers/routes.cjs');
module.exports = async ({ page, go, shot, fail }) => {
const stray = new Map();
for (const r of routes()) {
for (const r of staticRoutes()) {
await go(r);
await shot(r);
// Log lines are data, not copy.
+22
View File
@@ -50,6 +50,7 @@ import (
"github.com/samcharles93/archie-core/internal/infrastructure/controlplanestore"
"github.com/samcharles93/archie-core/internal/infrastructure/eventbus/nats"
"github.com/samcharles93/archie-core/internal/infrastructure/extension"
"github.com/samcharles93/archie-core/internal/infrastructure/issuerpc"
"github.com/samcharles93/archie-core/internal/infrastructure/kitrun"
"github.com/samcharles93/archie-core/internal/infrastructure/logrpc"
"github.com/samcharles93/archie-core/internal/infrastructure/messagingrpc"
@@ -587,9 +588,30 @@ func (b *boot) registerNATSRPC() error {
return err
}
b.addCleanup(unsubscribeTaskActions)
unsubscribeIssues, err := (&issuerpc.Server{Forge: b.issueForge, Allowed: b.managedRepository, Log: log}).Register(coreConn)
if err != nil {
log.Error("issue RPC register failed", "err", err)
return err
}
b.addCleanup(unsubscribeIssues)
return nil
}
// issueForge is the instance forge, or nil while none is configured.
func (b *boot) issueForge() forge.IssueForge {
if b.forgeClient == nil {
return nil
}
return b.forgeClient
}
// managedRepository refuses a repository the organisation has not added.
func (b *boot) managedRepository(ctx context.Context, owner, repo string) error {
_, err := b.repositoryBinding(ctx, owner, repo)
return err
}
func (b *boot) buildDaemon() {
log := b.log
b.d = &daemon.Daemon{
+5 -2
View File
@@ -19,8 +19,11 @@ const releaseUsage = `usage: archied release <command>
keygen -seed FILE write a new signing key's seed to FILE and print its public key
publish sign a release and push it to a release source's next channel
promote -version V point stable at a published release by hand
canary start point the canary at next and install a release on it
canary check judge the release on next and promote it or roll the canary back`
canary status show the canary's channel, installed release, window and verdict (read-only)
canary start point the canary at next and install a release on it, queued behind one being judged
canary check judge the release the canary runs: promote it, pick up a newer next, or roll back
canary promote promote the release the canary runs now, if its signals pass
canary unpin return the canary to the next channel from any pin`
// RunRelease is the publisher's side of a release source: it makes the key
// a source signs with and pushes signed releases to it.
+588 -92
View File
@@ -1,6 +1,7 @@
package archied
import (
"bytes"
"context"
"errors"
"flag"
@@ -8,6 +9,7 @@ import (
"io"
"os"
"os/exec"
"strings"
"time"
"oras.land/oras-go/v2/errdef"
@@ -25,136 +27,630 @@ const canaryInstallWait = 20 * time.Minute
const migrationsDir = "internal/infrastructure/postgres/migrations"
func runReleaseCanary(args []string, out, stderr io.Writer) int {
if len(args) == 0 || (args[0] != "start" && args[0] != "check") {
if len(args) == 0 {
fmt.Fprintln(stderr, releaseUsage)
return 2
}
flags := flag.NewFlagSet("release canary "+args[0], flag.ContinueOnError)
switch args[0] {
case "status":
return runCanaryStatus(args[1:], out, stderr)
case "start":
return runCanaryStart(args[1:], out, stderr)
case "check":
return runCanaryCheck(args[1:], out, stderr)
case "promote":
return runCanaryPromote(args[1:], out, stderr)
case "unpin":
return runCanaryUnpin(args[1:], out, stderr)
default:
fmt.Fprintln(stderr, releaseUsage)
return 2
}
}
// addCanaryFlags registers the flags every canary subcommand takes.
func addCanaryFlags(flags *flag.FlagSet) (dashboardURL, tokenEnv *string) {
dashboardURL = flags.String("dashboard", "", "canary dashboard URL")
tokenEnv = flags.String("token-env", "ARCHIE_CANARY_TOKEN", "environment variable holding a dashboard API token for the canary")
return dashboardURL, tokenEnv
}
// addSourceFlags registers the flags naming the release source.
func addSourceFlags(flags *flag.FlagSet) (reference, key *string) {
reference = flags.String("source", "", "OCI repository releases are published to; blank is the official releases")
key = flags.String("key", "", "base64 ed25519 public key the source signs with")
return reference, key
}
func canaryDashboard(url, tokenEnv string) (releaseupdate.CanaryDashboard, error) {
if url == "" {
return releaseupdate.CanaryDashboard{}, errors.New("-dashboard is required")
}
return releaseupdate.CanaryDashboard{URL: url, Token: os.Getenv(tokenEnv)}, nil
}
func canarySource(reference, key string) (distribution.Source, error) {
return distribution.Parse(reference, key, distribution.DefaultReleases())
}
// fetchChannel fetches a channel tag; found is false when the channel has no
// release yet.
func fetchChannel(ctx context.Context, source distribution.Source, tag string) (release releaseupdate.Release, found bool, err error) {
release, err = releaseupdate.Fetch(ctx, source, tag)
if errors.Is(err, errdef.ErrNotFound) {
return releaseupdate.Release{}, false, nil
}
return release, err == nil, err
}
// publishedChannels are the releases the source currently publishes.
type publishedChannels struct {
next, stable releaseupdate.Release
hasNext, hasStable bool
}
func fetchChannels(ctx context.Context, source distribution.Source) (publishedChannels, error) {
var channels publishedChannels
var err error
if channels.next, channels.hasNext, err = fetchChannel(ctx, source, "next"); err != nil {
return channels, err
}
if channels.stable, channels.hasStable, err = fetchChannel(ctx, source, "stable"); err != nil {
return channels, err
}
return channels, nil
}
// canaryRun is the release the canary actually runs: its version, when it was
// installed and its fetched release statement.
type canaryRun struct {
version string
installedAt time.Time
release releaseupdate.Release
found bool
}
// readCanaryRun reads what the canary runs. A host that doesn't record install
// times (it updated before they were tracked) degrades to the release
// statement's issue time.
func readCanaryRun(ctx context.Context, dashboard releaseupdate.CanaryDashboard, source distribution.Source) (canaryRun, error) {
var run canaryRun
version, installedAt, _, err := dashboard.Installed(ctx)
if err != nil {
return run, err
}
run.version, run.installedAt = version, installedAt
release, found, err := fetchChannel(ctx, source, releaseupdate.Tag("exact-pin", version))
if err != nil {
return run, err
}
run.release, run.found = release, found
if run.installedAt.IsZero() && found {
run.installedAt = release.Statement.IssuedAt
}
return run, nil
}
// newerIssued reports whether release a was issued after b. IssuedAt, not
// semver order: Promote refuses a release issued before the one stable names,
// so the judge must agree with that replay rule.
func newerIssued(a, b releaseupdate.Release) bool {
return a.Statement.IssuedAt.After(b.Statement.IssuedAt)
}
func displayVersion(version string) string {
if version == "" {
return "unknown"
}
return version
}
func displayRelease(release releaseupdate.Release, found bool) string {
if !found {
return "none"
}
return release.Statement.Version
}
func printVerdict(out io.Writer, version string, verdict releaseupdate.CanaryVerdict) {
fmt.Fprintf(out, "%s: %s\n", displayVersion(version), verdict.Decision)
for _, reason := range verdict.Reasons {
fmt.Fprintf(out, " %s\n", reason)
}
}
func runCanaryStatus(args []string, out, stderr io.Writer) int {
flags := flag.NewFlagSet("release canary status", flag.ContinueOnError)
flags.SetOutput(stderr)
dashboardURL := flags.String("dashboard", "", "canary dashboard URL")
tokenEnv := flags.String("token-env", "ARCHIE_CANARY_TOKEN", "environment variable holding a dashboard API token for the canary")
version := flags.String("version", "", "release to install (start)")
reference := flags.String("source", "", "OCI repository releases are published to; blank is the official releases (check)")
key := flags.String("key", "", "base64 ed25519 public key the source signs with (check)")
latest := flags.String("latest", "", "comma-separated image repositories whose v<version> tag becomes latest on promotion (check)")
checkout := flags.String("checkout", ".", "repository checkout with release tags, to see whether a release migrated the database (check)")
if err := flags.Parse(args[1:]); err != nil {
dashboardURL, tokenEnv := addCanaryFlags(flags)
reference, key := addSourceFlags(flags)
window := flags.Duration("window", releaseupdate.CanaryWindow, "window a release is judged over, counted from its install")
if err := flags.Parse(args); err != nil {
return 2
}
dashboard := releaseupdate.CanaryDashboard{URL: *dashboardURL, Token: os.Getenv(*tokenEnv)}
if *dashboardURL == "" {
fmt.Fprintln(stderr, "release canary: -dashboard is required")
return 2
}
ctx := context.Background()
if args[0] == "start" {
if *version == "" {
fmt.Fprintln(stderr, "release canary start: -version is required")
return 2
}
if err := startCanary(ctx, dashboard, *version); err != nil {
fmt.Fprintln(stderr, err)
return 1
}
fmt.Fprintf(out, "canary follows next and runs %s\n", *version)
return 0
}
source, err := distribution.Parse(*reference, *key, distribution.DefaultReleases())
dashboard, err := canaryDashboard(*dashboardURL, *tokenEnv)
if err != nil {
fmt.Fprintln(stderr, "release canary check:", err)
fmt.Fprintln(stderr, "release canary status:", err)
return 2
}
rolledBack, err := checkCanary(ctx, dashboard, source, splitList(*latest), *checkout, out)
source, err := canarySource(*reference, *key)
if err != nil {
fmt.Fprintln(stderr, err)
return 1
fmt.Fprintln(stderr, "release canary status:", err)
return 2
}
if rolledBack {
if err := printCanaryStatus(context.Background(), dashboard, source, *window, out); err != nil {
fmt.Fprintln(stderr, "release canary status:", err)
return 1
}
return 0
}
func startCanary(ctx context.Context, dashboard releaseupdate.CanaryDashboard, version string) error {
if err := dashboard.Follow(ctx, "next", ""); err != nil {
return fmt.Errorf("point the canary at next: %w", err)
}
return dashboard.Install(ctx, version, canaryInstallWait)
}
// checkCanary judges the release on the next channel once and acts on the
// result. It reports rolledBack so that the caller can fail the run.
func checkCanary(ctx context.Context, dashboard releaseupdate.CanaryDashboard, source distribution.Source, latest []string, checkout string, out io.Writer) (rolledBack bool, err error) {
next, err := releaseupdate.Fetch(ctx, source, "next")
if err != nil {
return false, err
}
stable, err := releaseupdate.Fetch(ctx, source, "stable")
hasStable := err == nil
if err != nil && !errors.Is(err, errdef.ErrNotFound) {
return false, err
}
release := next.Statement
if release.Prerelease || (hasStable && !release.IssuedAt.After(stable.Statement.IssuedAt)) {
fmt.Fprintf(out, "nothing to judge: next is %s and isn't a newer release than stable\n", release.Version)
return false, nil
}
// printCanaryStatus reports the canary's channel, installed release, the
// published channels, window progress and the verdict the current signals
// would give, without acting on anything.
func printCanaryStatus(ctx context.Context, dashboard releaseupdate.CanaryDashboard, source distribution.Source, window time.Duration, out io.Writer) error {
channel, pin, err := dashboard.Channel(ctx)
if err != nil {
return false, err
return err
}
if channel != "next" {
fmt.Fprintf(out, "nothing to judge: the canary is held on %s %s\n", channel, pin)
return false, nil
channels, err := fetchChannels(ctx, source)
if err != nil {
return err
}
run, err := readCanaryRun(ctx, dashboard, source)
if err != nil {
return err
}
printCanaryState(out, channel, pin, run, channels, window)
return printCanaryVerdict(ctx, dashboard, run, window, out)
}
func printCanaryState(out io.Writer, channel, pin string, run canaryRun, channels publishedChannels, window time.Duration) {
fmt.Fprintf(out, "channel: %s\n", channelDisplay(channel, pin))
fmt.Fprintf(out, "installed: %s", displayVersion(run.version))
if run.installedAt.IsZero() {
fmt.Fprintln(out, " (installed_at not reported)")
} else {
fmt.Fprintf(out, " since %s\n", run.installedAt.Format(time.RFC3339))
}
fmt.Fprintf(out, "stable: %s\n", displayRelease(channels.stable, channels.hasStable))
fmt.Fprintf(out, "next: %s\n", displayRelease(channels.next, channels.hasNext))
if run.installedAt.IsZero() {
fmt.Fprintln(out, "window: unknown, no install time to count from")
} else {
elapsed := time.Since(run.installedAt)
fmt.Fprintf(out, "window: %s of %s elapsed since the install\n", elapsed.Round(time.Minute), window)
}
}
func printCanaryVerdict(ctx context.Context, dashboard releaseupdate.CanaryDashboard, run canaryRun, window time.Duration, out io.Writer) error {
if run.installedAt.IsZero() {
fmt.Fprintln(out, "verdict: unknown, no install time to judge from")
return nil
}
now := time.Now().UTC()
signals, err := dashboard.Signals(ctx, release.IssuedAt, now)
signals, err := dashboard.Signals(ctx, run.installedAt, now, window)
if err != nil {
return false, fmt.Errorf("read canary signals: %w", err)
return fmt.Errorf("read canary signals: %w", err)
}
verdict := releaseupdate.JudgeCanary(release.Version, now.Sub(release.IssuedAt), signals)
fmt.Fprintf(out, "%s: %s\n", release.Version, verdict.Decision)
printSignals(out, signals)
verdict := releaseupdate.JudgeCanary(run.version, run.installedAt, now, window, signals)
fmt.Fprintf(out, "verdict on %s: %s (read-only)\n", displayVersion(run.version), verdict.Decision)
for _, reason := range verdict.Reasons {
fmt.Fprintf(out, " %s\n", reason)
}
switch verdict.Decision {
case releaseupdate.CanaryPromote:
if _, err := releaseupdate.Promote(ctx, source.Reference, source.Key, release.Version, latest); err != nil {
return false, fmt.Errorf("promote %s: %w", release.Version, err)
return nil
}
func printSignals(out io.Writer, s releaseupdate.CanarySignals) {
fmt.Fprintf(out, "signals: runs %q, ready=%t, new action alerts=%d\n", s.Installed, s.Ready, len(s.NewActionAlerts))
fmt.Fprintf(out, " tasks finished %d, parked or died %d (baseline %d, %d)\n",
s.Canary.Finished, s.Canary.Bad, s.Baseline.Finished, s.Baseline.Bad)
fmt.Fprintf(out, " error lines %d, log covers %s (baseline %d, covers %s)\n",
s.Canary.Errors, s.Canary.Covered.Round(time.Minute), s.Baseline.Errors, s.Baseline.Covered.Round(time.Minute))
}
func channelDisplay(channel, pin string) string {
if pin != "" {
return channel + " (pinned to " + pin + ")"
}
return channel
}
func runCanaryStart(args []string, out, stderr io.Writer) int {
flags := flag.NewFlagSet("release canary start", flag.ContinueOnError)
flags.SetOutput(stderr)
dashboardURL, tokenEnv := addCanaryFlags(flags)
reference, key := addSourceFlags(flags)
version := flags.String("version", "", "release to install on the canary")
force := flags.Bool("force", false, "install even when the canary is still judging a newer release")
if err := flags.Parse(args); err != nil {
return 2
}
dashboard, err := canaryDashboard(*dashboardURL, *tokenEnv)
if err != nil {
fmt.Fprintln(stderr, "release canary start:", err)
return 2
}
if *version == "" {
fmt.Fprintln(stderr, "release canary start: -version is required")
return 2
}
source, err := canarySource(*reference, *key)
if err != nil {
fmt.Fprintln(stderr, "release canary start:", err)
return 2
}
if err := startCanary(context.Background(), dashboard, source, *version, *force, out); err != nil {
fmt.Fprintln(stderr, "release canary start:", err)
return 1
}
return 0
}
// startCanary points the canary at next — clearing any rollback pin — and
// installs version on it. A release published while the canary still judges a
// newer one queues behind it, so the running window isn't restarted under a
// different release.
func startCanary(ctx context.Context, dashboard releaseupdate.CanaryDashboard, source distribution.Source, version string, force bool, out io.Writer) error {
// The channel is read before following next: a canary already on next is
// judging the release it runs, and a new release queues behind that one.
channel, _, err := dashboard.Channel(ctx)
if err != nil {
return err
}
if err := dashboard.Follow(ctx, "next", ""); err != nil {
return fmt.Errorf("point the canary at next: %w", err)
}
if !force {
queued, err := queuedBehind(ctx, dashboard, source, channel, version, out)
if err != nil {
return err
}
fmt.Fprintf(out, "stable is %s\n", release.Version)
if queued {
return nil
}
}
if err := dashboard.Install(ctx, version, canaryInstallWait); err != nil {
return err
}
fmt.Fprintf(out, "the canary follows next and runs %s\n", version)
return nil
}
// queuedBehind reports whether version must wait: the canary already follows
// next and is still judging the release it runs (a judgeable one), which is a
// different release from version.
func queuedBehind(ctx context.Context, dashboard releaseupdate.CanaryDashboard, source distribution.Source, channel, version string, out io.Writer) (bool, error) {
if channel != "next" {
return false, nil
case releaseupdate.CanaryRollBack:
return true, rollBackCanary(ctx, dashboard, release.Version, stable.Statement.Version, hasStable, checkout, out)
}
run, err := readCanaryRun(ctx, dashboard, source)
if err != nil {
return false, err
}
if run.version == version {
return false, nil
}
channels, err := fetchChannels(ctx, source)
if err != nil {
return false, err
}
if judgeable(run, channels) {
fmt.Fprintf(out, "%s is queued behind %s, which the canary is still judging\n", version, run.version)
return true, nil
}
return false, nil
}
// rollBackCanary pins the canary to the stable release and reinstalls it.
// Going back past a migration needs the pre-update database snapshot, which
// loses every write since, so a release that migrated is held instead.
func rollBackCanary(ctx context.Context, dashboard releaseupdate.CanaryDashboard, failed, stable string, hasStable bool, checkout string, out io.Writer) error {
if !hasStable || migrated(ctx, checkout, stable, failed) {
if err := dashboard.Follow(ctx, "exact-pin", failed); err != nil {
return fmt.Errorf("hold the canary on %s: %w", failed, err)
func runCanaryCheck(args []string, out, stderr io.Writer) int {
flags := flag.NewFlagSet("release canary check", flag.ContinueOnError)
flags.SetOutput(stderr)
dashboardURL, tokenEnv := addCanaryFlags(flags)
reference, key := addSourceFlags(flags)
latest := flags.String("latest", "", "comma-separated image repositories whose v<version> tag becomes latest on promotion")
checkout := flags.String("checkout", ".", "repository checkout with release tags, to see whether a release migrated the database")
window := flags.Duration("window", releaseupdate.CanaryWindow, "window the installed release is judged over, counted from its install")
if err := flags.Parse(args); err != nil {
return 2
}
dashboard, err := canaryDashboard(*dashboardURL, *tokenEnv)
if err != nil {
fmt.Fprintln(stderr, "release canary check:", err)
return 2
}
source, err := canarySource(*reference, *key)
if err != nil {
fmt.Fprintln(stderr, "release canary check:", err)
return 2
}
code, err := checkCanary(context.Background(), dashboard, source, splitList(*latest), *checkout, *window, out)
if err != nil {
fmt.Fprintln(stderr, "release canary check:", err)
return 1
}
return code
}
// checkCanary judges the release the canary actually runs — not the moving
// next tag — and acts on the verdict. The exit code is 1 when the run should
// show red: a rollback happened, or the canary is held on a pin that can no
// longer be installed.
func checkCanary(ctx context.Context, dashboard releaseupdate.CanaryDashboard, source distribution.Source, latest []string, checkout string, window time.Duration, out io.Writer) (int, error) {
channels, err := fetchChannels(ctx, source)
if err != nil {
return 1, err
}
channel, pin, err := dashboard.Channel(ctx)
if err != nil {
return 1, err
}
if channel != "next" {
return heldCanary(ctx, source, channel, pin, out)
}
run, err := readCanaryRun(ctx, dashboard, source)
if err != nil {
return 1, err
}
if !judgeable(run, channels) {
fmt.Fprintf(out, "nothing to judge: the canary runs %s\n", displayVersion(run.version))
return pickupNext(ctx, dashboard, channels, run.version, out)
}
return judgeInstalled(ctx, dashboard, source, run, channels, latest, checkout, window, out)
}
// judgeable reports whether the installed release is one the canary judges: a
// published, non-prerelease release issued after stable's.
func judgeable(run canaryRun, channels publishedChannels) bool {
if !run.found || run.release.Statement.Prerelease {
return false
}
return !channels.hasStable || newerIssued(run.release, channels.stable)
}
func judgeInstalled(ctx context.Context, dashboard releaseupdate.CanaryDashboard, source distribution.Source, run canaryRun, channels publishedChannels, latest []string, checkout string, window time.Duration, out io.Writer) (int, error) {
now := time.Now().UTC()
signals, err := dashboard.Signals(ctx, run.installedAt, now, window)
if err != nil {
return 1, fmt.Errorf("read canary signals: %w", err)
}
verdict := releaseupdate.JudgeCanary(run.version, run.installedAt, now, window, signals)
printVerdict(out, run.version, verdict)
switch verdict.Decision {
case releaseupdate.CanaryPromote:
return 0, promoteJudged(ctx, dashboard, source, run, channels, latest, out)
case releaseupdate.CanaryRollBack:
return 1, rollBackCanary(ctx, dashboard, source, run, channels, checkout, out)
default:
return 0, nil
}
}
// promoteJudged moves stable to the judged release, then picks up a newer next
// queued behind it.
func promoteJudged(ctx context.Context, dashboard releaseupdate.CanaryDashboard, source distribution.Source, run canaryRun, channels publishedChannels, latest []string, out io.Writer) error {
if _, err := releaseupdate.Promote(ctx, source.Reference, source.Key, run.version, latest); err != nil {
return fmt.Errorf("promote %s: %w", run.version, err)
}
fmt.Fprintf(out, "stable is %s\n", run.version)
fresh := channels
fresh.stable, fresh.hasStable = run.release, true
_, err := pickupNext(ctx, dashboard, fresh, run.version, out)
return err
}
// pickupNext installs the next release when it is newer than stable and the
// canary doesn't already run it.
func pickupNext(ctx context.Context, dashboard releaseupdate.CanaryDashboard, channels publishedChannels, installed string, out io.Writer) (int, error) {
if !channels.hasNext || channels.next.Statement.Version == installed {
return 0, nil
}
if channels.hasStable && !newerIssued(channels.next, channels.stable) {
return 0, nil
}
if err := dashboard.Install(ctx, channels.next.Statement.Version, canaryInstallWait); err != nil {
return 1, err
}
fmt.Fprintf(out, "the canary picks up %s from next\n", channels.next.Statement.Version)
return 0, nil
}
// heldCanary reports a canary that isn't following next. Its pin must still be
// installable: a stranded pin fails the run instead of looking green.
func heldCanary(ctx context.Context, source distribution.Source, channel, pin string, out io.Writer) (int, error) {
if pin != "" {
err := checkPin(ctx, source, pin)
if errors.Is(err, errStrandedPin) {
fmt.Fprintf(out, "ERROR: the canary is held on %s %s, which is stranded: %v; unpin it or republish the release\n", channel, pin, err)
return 1, nil
}
fmt.Fprintf(out, "%s changed the database schema or there's no stable release to return to: the canary is held on %s and stable is unchanged; fix forward\n", failed, failed)
return nil
if err != nil {
return 1, err
}
fmt.Fprintf(out, "held on %s %s, nothing to judge\n", channel, pin)
return 0, nil
}
if err := dashboard.Follow(ctx, "exact-pin", stable); err != nil {
return fmt.Errorf("pin the canary to %s: %w", stable, err)
fmt.Fprintf(out, "held on %s, nothing to judge\n", channel)
return 0, nil
}
// errStrandedPin marks a held pin that can no longer be installed.
var errStrandedPin = errors.New("pin is stranded")
// checkPin fails when a held pin can no longer be installed: the release is
// gone from the source, or its artifacts are.
func checkPin(ctx context.Context, source distribution.Source, pin string) error {
release, found, err := fetchChannel(ctx, source, releaseupdate.Tag("exact-pin", pin))
if err != nil {
return err
}
if err := dashboard.Install(ctx, stable, canaryInstallWait); err != nil {
return fmt.Errorf("return the canary to %s: %w", stable, err)
if !found {
return fmt.Errorf("%w: no release %s in the source", errStrandedPin, releaseupdate.Tag("exact-pin", pin))
}
if err := releaseupdate.Installable(ctx, source, release); err != nil {
return fmt.Errorf("%w: %w", errStrandedPin, err)
}
fmt.Fprintf(out, "the canary is back on %s and stable is unchanged\n", stable)
return nil
}
// rollBackCanary returns the canary to stable after a failed verdict. stable
// is checked installable before the canary is pinned to it: a pin whose
// artifacts are gone would strand the canary, so the run fails loudly and the
// canary stays on next. A release that changed the migrations is held on the
// failed version instead, because going back past a migration needs the
// pre-update database snapshot.
func rollBackCanary(ctx context.Context, dashboard releaseupdate.CanaryDashboard, source distribution.Source, run canaryRun, channels publishedChannels, checkout string, out io.Writer) error {
failed := run.version
if !channels.hasStable {
return holdCanary(ctx, dashboard, failed, "there's no stable release to return to", out)
}
stable := channels.stable
if err := releaseupdate.Installable(ctx, source, stable); err != nil {
return fmt.Errorf("stable %s is not installable, the canary stays on next: %w", stable.Statement.Version, err)
}
changed, err := migrated(ctx, checkout, stable.Statement.Version, failed)
if err != nil {
return holdCanary(ctx, dashboard, failed, err.Error(), out)
}
if changed {
return holdCanary(ctx, dashboard, failed, failed+" changed the database schema", out)
}
if err := dashboard.Follow(ctx, "exact-pin", stable.Statement.Version); err != nil {
return fmt.Errorf("pin the canary to %s: %w", stable.Statement.Version, err)
}
if err := dashboard.Install(ctx, stable.Statement.Version, canaryInstallWait); err != nil {
return fmt.Errorf("return the canary to %s: %w", stable.Statement.Version, err)
}
fmt.Fprintf(out, "the canary is back on %s and stable is unchanged\n", stable.Statement.Version)
return nil
}
func holdCanary(ctx context.Context, dashboard releaseupdate.CanaryDashboard, version, why string, out io.Writer) error {
if err := dashboard.Follow(ctx, "exact-pin", version); err != nil {
return fmt.Errorf("hold the canary on %s: %w", version, err)
}
fmt.Fprintf(out, "%s: the canary is held on %s and stable is unchanged; fix forward\n", why, version)
return nil
}
// runGitDiff runs git diff --quiet between two release tags and reports its
// exit code and stderr. It is a variable so tests can drive the three
// outcomes without a repository.
var runGitDiff = func(ctx context.Context, checkout, from, to string) (int, string, error) {
cmd := exec.CommandContext(ctx, "git", "-C", checkout, "diff", "--quiet", "v"+from, "v"+to, "--", migrationsDir)
var buf bytes.Buffer
cmd.Stderr = &buf
err := cmd.Run()
if err == nil {
return 0, "", nil
}
if exitErr, ok := errors.AsType[*exec.ExitError](err); ok {
return exitErr.ExitCode(), buf.String(), err
}
return -1, buf.String(), err
}
// migrated reports whether the migrations differ between two release tags.
// If it can't tell, it says they differ, so the canary is held rather than
// rolled back.
func migrated(ctx context.Context, checkout, from, to string) bool {
err := exec.CommandContext(ctx, "git", "-C", checkout, "diff", "--quiet", "v"+from, "v"+to, "--", migrationsDir).Run()
return err != nil
// It returns true, nil when they differ, false, nil when they are identical,
// and true, err when git could not answer, with the reason in err. Every
// non-identical outcome holds the canary rather than rolls it back.
func migrated(ctx context.Context, checkout, from, to string) (bool, error) {
code, stderr, runErr := runGitDiff(ctx, checkout, from, to)
switch code {
case 0:
return false, nil
case 1:
return true, nil
}
reason := strings.TrimSpace(stderr)
if reason == "" && runErr != nil {
reason = runErr.Error()
}
return true, fmt.Errorf("could not tell whether %s changed the database schema: %s", to, reason)
}
func runCanaryPromote(args []string, out, stderr io.Writer) int {
flags := flag.NewFlagSet("release canary promote", flag.ContinueOnError)
flags.SetOutput(stderr)
dashboardURL, tokenEnv := addCanaryFlags(flags)
reference, key := addSourceFlags(flags)
latest := flags.String("latest", "", "comma-separated image repositories whose v<version> tag becomes latest on promotion")
version := flags.String("version", "", "release expected on the canary; must match what it runs")
window := flags.Duration("window", releaseupdate.CanaryWindow, "window the signals are read over, counted from the install")
if err := flags.Parse(args); err != nil {
return 2
}
dashboard, err := canaryDashboard(*dashboardURL, *tokenEnv)
if err != nil {
fmt.Fprintln(stderr, "release canary promote:", err)
return 2
}
source, err := canarySource(*reference, *key)
if err != nil {
fmt.Fprintln(stderr, "release canary promote:", err)
return 2
}
code, err := promoteCanaryNow(context.Background(), dashboard, source, *version, splitList(*latest), *window, out)
if err != nil {
fmt.Fprintln(stderr, "release canary promote:", err)
return 1
}
return code
}
// promoteCanaryNow promotes the release the canary runs right now, without
// waiting out the window: the signals must show no hard failure and no
// relative breach. It refuses, with exit 1, when anything fails.
func promoteCanaryNow(ctx context.Context, dashboard releaseupdate.CanaryDashboard, source distribution.Source, expectVersion string, latest []string, window time.Duration, out io.Writer) (int, error) {
run, err := readCanaryRun(ctx, dashboard, source)
if err != nil {
return 1, err
}
channel, pin, err := dashboard.Channel(ctx)
if err != nil {
return 1, err
}
fmt.Fprintf(out, "channel: %s\n", channelDisplay(channel, pin))
if expectVersion != "" && expectVersion != run.version {
return 1, fmt.Errorf("the canary runs %s, not %s", displayVersion(run.version), expectVersion)
}
if !run.found {
return 1, fmt.Errorf("no release %s in the source to promote", displayVersion(run.version))
}
now := time.Now().UTC()
signals, err := dashboard.Signals(ctx, run.installedAt, now, window)
if err != nil {
return 1, fmt.Errorf("read canary signals: %w", err)
}
verdict := releaseupdate.JudgeCanaryNow(run.version, now, window, signals)
printVerdict(out, run.version, verdict)
if verdict.Decision == releaseupdate.CanaryRollBack {
fmt.Fprintln(out, "refusing to promote: the canary's signals fail")
return 1, nil
}
if _, err := releaseupdate.Promote(ctx, source.Reference, source.Key, run.version, latest); err != nil {
return 1, fmt.Errorf("promote %s: %w", run.version, err)
}
fmt.Fprintf(out, "stable is %s\n", run.version)
return 0, nil
}
func runCanaryUnpin(args []string, out, stderr io.Writer) int {
flags := flag.NewFlagSet("release canary unpin", flag.ContinueOnError)
flags.SetOutput(stderr)
dashboardURL, tokenEnv := addCanaryFlags(flags)
if err := flags.Parse(args); err != nil {
return 2
}
dashboard, err := canaryDashboard(*dashboardURL, *tokenEnv)
if err != nil {
fmt.Fprintln(stderr, "release canary unpin:", err)
return 2
}
if err := dashboard.Follow(context.Background(), "next", ""); err != nil {
fmt.Fprintln(stderr, "release canary unpin:", err)
return 1
}
fmt.Fprintln(out, "the canary follows next again, with no pin")
return 0
}
@@ -0,0 +1,57 @@
package archied
import (
"context"
"strings"
"testing"
)
func TestMigrated(t *testing.T) {
tests := []struct {
name string
code int
stderr string
wantChanged bool
wantErr bool
wantSubstr string
}{
{
name: "identical migrations",
code: 0,
wantChanged: false,
},
{
name: "migrations differ",
code: 1,
wantChanged: true,
},
{
name: "git cannot answer",
code: 128,
stderr: "fatal: bad revision 'v1.0.0'\n",
wantChanged: true,
wantErr: true,
wantSubstr: "fatal: bad revision 'v1.0.0'",
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
old := runGitDiff
runGitDiff = func(context.Context, string, string, string) (int, string, error) {
return tt.code, tt.stderr, nil
}
t.Cleanup(func() { runGitDiff = old })
changed, err := migrated(context.Background(), ".", "1.0.0", "1.1.0")
if changed != tt.wantChanged {
t.Fatalf("changed = %v, want %v", changed, tt.wantChanged)
}
if (err != nil) != tt.wantErr {
t.Fatalf("err = %v, want err %v", err, tt.wantErr)
}
if tt.wantErr && !strings.Contains(err.Error(), tt.wantSubstr) {
t.Fatalf("err = %q, want substring %q", err, tt.wantSubstr)
}
})
}
}
+7
View File
@@ -77,6 +77,13 @@ func (b *boot) removeTaskWorktree(task *taskactions.Task) error {
if b.d == nil {
return nil
}
// A run that handed off shares the work item's worktree with its
// successor; archiving the old run must not remove it.
if current, err := b.stateStore.TaskByIssue(context.Background(), task.Owner, task.Repo, task.IssueNumber); err != nil {
return err
} else if current != nil && current.ID != task.ID {
return nil
}
return b.d.RemoveWorktree(&workflow.Task{Owner: task.Owner, Repo: task.Repo, Identity: task.Identity, IssueNumber: task.IssueNumber, Org: task.Org})
}
+1 -4
View File
@@ -295,10 +295,7 @@ func hasConfigSection(document map[string]any, section string) bool {
}
func updateStateDir() string {
if dir := os.Getenv("ARCHIE_UPDATE_STATE_DIR"); dir != "" {
return dir
}
return configuration.DefaultUpdateStateDir()
return configuration.UpdateStateDir()
}
// resumeUpdate hands an interrupted transaction to the frozen watchdog and
+39 -1
View File
@@ -12,7 +12,9 @@ import (
"github.com/samcharles93/archie-core/internal/agentexec/modelloop"
"github.com/samcharles93/archie-core/internal/config"
"github.com/samcharles93/archie-core/internal/domain/access"
"github.com/samcharles93/archie-core/internal/domain/curator"
"github.com/samcharles93/archie-core/internal/domain/org"
"github.com/samcharles93/archie-core/internal/domain/usage"
"github.com/samcharles93/archie-core/internal/events"
"github.com/samcharles93/archie-core/internal/infrastructure/controlplanestore"
@@ -40,7 +42,7 @@ func (b *server) setupCurators(ctx context.Context) bool {
MemoryEngines: b.memEngines,
Skills: skillcurator.NewStore(skillsRoot),
Conversations: sessioncurator.NewAdapter(b.chatSessionStore, b.cfg.BotUser),
LLM: curatorLLMRunner{llm: b.chatLLM, outcomes: b.providerOutcomes, usage: b.usage},
LLM: actingPrincipalRunner{inner: curatorLLMRunner{llm: b.chatLLM, outcomes: b.providerOutcomes, usage: b.usage}, resolve: b.actingPrincipal},
Model: model,
})
@@ -95,6 +97,42 @@ func (s curatorEventSink) Emit(kind, detail string, data map[string]any) {
s.b.Publish(context.Background(), events.Event{Kind: kind, Detail: detail, Data: data})
}
// actingPrincipal resolves the identity the instance acts as for automated
// work (the default chat identity, falling back to the bot user), so curator
// passes authorize guarded tools exactly as that identity would.
func (b *server) actingPrincipal(ctx context.Context) (access.Principal, error) {
name := b.defaultChatIdentity
if name == "" {
name = b.cfg.BotUser
}
if name == "" {
return access.Principal{}, fmt.Errorf("no acting identity configured (chat identity or bot user)")
}
idn, err := b.stateStore.ResolveLegacyName(ctx, name)
if err != nil {
return access.Principal{}, fmt.Errorf("resolve acting identity %q: %w", name, err)
}
return b.stateStore.PrincipalFor(ctx, idn.ID)
}
// actingPrincipalRunner attaches the acting principal to every curator model
// call. Tool handlers run in the call's context, so guarded tools see the
// curator as the instance's identity, never as an anonymous or caller-supplied
// one; a pass that cannot resolve the principal fails closed.
type actingPrincipalRunner struct {
inner curator.LLMRunner
resolve func(context.Context) (access.Principal, error)
}
func (r actingPrincipalRunner) Chat(ctx context.Context, req curator.ChatRequest) (curator.ChatResult, error) {
p, err := r.resolve(ctx)
if err != nil {
return curator.ChatResult{}, fmt.Errorf("curator pass: %w", err)
}
ctx = org.WithScope(org.WithOrg(access.WithPrincipal(ctx, p), p.Org), p.Org)
return r.inner.Chat(ctx, req)
}
// curatorLLMRunner adapts the shared ai-sdk runtime to the curator
// family's narrow LLMRunner contract: one model reference, plain
// messages, and the declared tool set -- no streaming. Tools are built
+9
View File
@@ -18,6 +18,7 @@ import (
"github.com/samcharles93/archie-core/internal/gateway"
infraaccess "github.com/samcharles93/archie-core/internal/infrastructure/access"
"github.com/samcharles93/archie-core/internal/infrastructure/controlplanerpc"
"github.com/samcharles93/archie-core/internal/infrastructure/issuerpc"
"github.com/samcharles93/archie-core/internal/infrastructure/staterpc"
"github.com/samcharles93/archie-core/internal/infrastructure/taskmcp"
"github.com/samcharles93/archie-core/internal/plugin"
@@ -109,6 +110,14 @@ func (b *server) registerWorkflowTools(ctx context.Context) {
authz := infraaccess.StartLive(ctx, b.stateStore.ListPolicies, b.stateStore.PoliciesVersion, b.log)
workflows := controlplanerpc.NewWorkflows(b.stateStore.ControlPlane())
entries := append(gateway.WorkflowTools(workflows, authz), gateway.IntegrationTools(integrationSource{Workflows: workflows, store: b.stateStore}, authz)...)
// Run-inspection tools back long-running curators (self-intake first);
// the State Store client satisfies the source directly.
entries = append(entries, gateway.RunTools(b.stateStore, authz)...)
// Issue tools reach the forge through the daemon, which alone holds its
// credentials; without the broker connection they are not registered.
if b.taskActionsConn != nil {
entries = append(entries, gateway.IssueTools(issuerpc.Client{Conn: b.taskActionsConn, Timeout: 30 * time.Second}, authz)...)
}
for _, entry := range entries {
if err := b.toolReg.Register(entry); err != nil {
b.log.WarnContext(ctx, "workflow tool registration failed", "tool", entry.Name, "err", err)
File diff suppressed because it is too large. Load diff
@@ -23,6 +23,7 @@ const (
ForgeService_AssignedIssues_FullMethodName = "/forge.v1.ForgeService/AssignedIssues"
ForgeService_IssuesWithLabel_FullMethodName = "/forge.v1.ForgeService/IssuesWithLabel"
ForgeService_Comment_FullMethodName = "/forge.v1.ForgeService/Comment"
ForgeService_CreateIssue_FullMethodName = "/forge.v1.ForgeService/CreateIssue"
ForgeService_CloseIssue_FullMethodName = "/forge.v1.ForgeService/CloseIssue"
ForgeService_React_FullMethodName = "/forge.v1.ForgeService/React"
ForgeService_SetStateLabel_FullMethodName = "/forge.v1.ForgeService/SetStateLabel"
@@ -60,6 +61,7 @@ type ForgeServiceClient interface {
AssignedIssues(ctx context.Context, in *AssignedIssuesRequest, opts ...grpc.CallOption) (*AssignedIssuesResponse, error)
IssuesWithLabel(ctx context.Context, in *IssuesWithLabelRequest, opts ...grpc.CallOption) (*IssuesWithLabelResponse, error)
Comment(ctx context.Context, in *CommentRequest, opts ...grpc.CallOption) (*CommentResponse, error)
CreateIssue(ctx context.Context, in *CreateIssueRequest, opts ...grpc.CallOption) (*CreateIssueResponse, error)
CloseIssue(ctx context.Context, in *CloseIssueRequest, opts ...grpc.CallOption) (*CloseIssueResponse, error)
React(ctx context.Context, in *ReactRequest, opts ...grpc.CallOption) (*ReactResponse, error)
SetStateLabel(ctx context.Context, in *SetStateLabelRequest, opts ...grpc.CallOption) (*SetStateLabelResponse, error)
@@ -132,6 +134,16 @@ func (c *forgeServiceClient) Comment(ctx context.Context, in *CommentRequest, op
return out, nil
}
func (c *forgeServiceClient) CreateIssue(ctx context.Context, in *CreateIssueRequest, opts ...grpc.CallOption) (*CreateIssueResponse, error) {
cOpts := append([]grpc.CallOption{grpc.StaticMethod()}, opts...)
out := new(CreateIssueResponse)
err := c.cc.Invoke(ctx, ForgeService_CreateIssue_FullMethodName, in, out, cOpts...)
if err != nil {
return nil, err
}
return out, nil
}
func (c *forgeServiceClient) CloseIssue(ctx context.Context, in *CloseIssueRequest, opts ...grpc.CallOption) (*CloseIssueResponse, error) {
cOpts := append([]grpc.CallOption{grpc.StaticMethod()}, opts...)
out := new(CloseIssueResponse)
@@ -338,6 +350,7 @@ type ForgeServiceServer interface {
AssignedIssues(context.Context, *AssignedIssuesRequest) (*AssignedIssuesResponse, error)
IssuesWithLabel(context.Context, *IssuesWithLabelRequest) (*IssuesWithLabelResponse, error)
Comment(context.Context, *CommentRequest) (*CommentResponse, error)
CreateIssue(context.Context, *CreateIssueRequest) (*CreateIssueResponse, error)
CloseIssue(context.Context, *CloseIssueRequest) (*CloseIssueResponse, error)
React(context.Context, *ReactRequest) (*ReactResponse, error)
SetStateLabel(context.Context, *SetStateLabelRequest) (*SetStateLabelResponse, error)
@@ -382,6 +395,9 @@ func (UnimplementedForgeServiceServer) IssuesWithLabel(context.Context, *IssuesW
func (UnimplementedForgeServiceServer) Comment(context.Context, *CommentRequest) (*CommentResponse, error) {
return nil, status.Error(codes.Unimplemented, "method Comment not implemented")
}
func (UnimplementedForgeServiceServer) CreateIssue(context.Context, *CreateIssueRequest) (*CreateIssueResponse, error) {
return nil, status.Error(codes.Unimplemented, "method CreateIssue not implemented")
}
func (UnimplementedForgeServiceServer) CloseIssue(context.Context, *CloseIssueRequest) (*CloseIssueResponse, error) {
return nil, status.Error(codes.Unimplemented, "method CloseIssue not implemented")
}
@@ -529,6 +545,24 @@ func _ForgeService_Comment_Handler(srv interface{}, ctx context.Context, dec fun
return interceptor(ctx, in, info, handler)
}
func _ForgeService_CreateIssue_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
in := new(CreateIssueRequest)
if err := dec(in); err != nil {
return nil, err
}
if interceptor == nil {
return srv.(ForgeServiceServer).CreateIssue(ctx, in)
}
info := &grpc.UnaryServerInfo{
Server: srv,
FullMethod: ForgeService_CreateIssue_FullMethodName,
}
handler := func(ctx context.Context, req interface{}) (interface{}, error) {
return srv.(ForgeServiceServer).CreateIssue(ctx, req.(*CreateIssueRequest))
}
return interceptor(ctx, in, info, handler)
}
func _ForgeService_CloseIssue_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
in := new(CloseIssueRequest)
if err := dec(in); err != nil {
@@ -869,6 +903,10 @@ var ForgeService_ServiceDesc = grpc.ServiceDesc{
MethodName: "Comment",
Handler: _ForgeService_Comment_Handler,
},
{
MethodName: "CreateIssue",
Handler: _ForgeService_CreateIssue_Handler,
},
{
MethodName: "CloseIssue",
Handler: _ForgeService_CloseIssue_Handler,
File diff suppressed because it is too large. Load diff
@@ -38,6 +38,7 @@ const (
AccountService_AddPersonalToken_FullMethodName = "/state.v1.AccountService/AddPersonalToken"
AccountService_ListPersonalTokens_FullMethodName = "/state.v1.AccountService/ListPersonalTokens"
AccountService_RevokePersonalToken_FullMethodName = "/state.v1.AccountService/RevokePersonalToken"
AccountService_BindIdentityPersonalToken_FullMethodName = "/state.v1.AccountService/BindIdentityPersonalToken"
)
// AccountServiceClient is the client API for AccountService service.
@@ -65,6 +66,12 @@ type AccountServiceClient interface {
AddPersonalToken(ctx context.Context, in *AddPersonalTokenRequest, opts ...grpc.CallOption) (*AddPersonalTokenResponse, error)
ListPersonalTokens(ctx context.Context, in *ListPersonalTokensRequest, opts ...grpc.CallOption) (*ListPersonalTokensResponse, error)
RevokePersonalToken(ctx context.Context, in *RevokePersonalTokenRequest, opts ...grpc.CallOption) (*RevokePersonalTokenResponse, error)
// BindIdentityPersonalToken binds a personal token to a NAMED identity
// instead of the caller's principal, which is how a service account gets
// its bearer credential. Only non-user targets are accepted: a person's
// token comes from AddPersonalToken. The audit actor is the caller's
// principal, never a request field.
BindIdentityPersonalToken(ctx context.Context, in *BindIdentityPersonalTokenRequest, opts ...grpc.CallOption) (*BindIdentityPersonalTokenResponse, error)
}
type accountServiceClient struct {
@@ -265,6 +272,16 @@ func (c *accountServiceClient) RevokePersonalToken(ctx context.Context, in *Revo
return out, nil
}
func (c *accountServiceClient) BindIdentityPersonalToken(ctx context.Context, in *BindIdentityPersonalTokenRequest, opts ...grpc.CallOption) (*BindIdentityPersonalTokenResponse, error) {
cOpts := append([]grpc.CallOption{grpc.StaticMethod()}, opts...)
out := new(BindIdentityPersonalTokenResponse)
err := c.cc.Invoke(ctx, AccountService_BindIdentityPersonalToken_FullMethodName, in, out, cOpts...)
if err != nil {
return nil, err
}
return out, nil
}
// AccountServiceServer is the server API for AccountService service.
// All implementations must embed UnimplementedAccountServiceServer
// for forward compatibility.
@@ -290,6 +307,12 @@ type AccountServiceServer interface {
AddPersonalToken(context.Context, *AddPersonalTokenRequest) (*AddPersonalTokenResponse, error)
ListPersonalTokens(context.Context, *ListPersonalTokensRequest) (*ListPersonalTokensResponse, error)
RevokePersonalToken(context.Context, *RevokePersonalTokenRequest) (*RevokePersonalTokenResponse, error)
// BindIdentityPersonalToken binds a personal token to a NAMED identity
// instead of the caller's principal, which is how a service account gets
// its bearer credential. Only non-user targets are accepted: a person's
// token comes from AddPersonalToken. The audit actor is the caller's
// principal, never a request field.
BindIdentityPersonalToken(context.Context, *BindIdentityPersonalTokenRequest) (*BindIdentityPersonalTokenResponse, error)
mustEmbedUnimplementedAccountServiceServer()
}
@@ -357,6 +380,9 @@ func (UnimplementedAccountServiceServer) ListPersonalTokens(context.Context, *Li
func (UnimplementedAccountServiceServer) RevokePersonalToken(context.Context, *RevokePersonalTokenRequest) (*RevokePersonalTokenResponse, error) {
return nil, status.Error(codes.Unimplemented, "method RevokePersonalToken not implemented")
}
func (UnimplementedAccountServiceServer) BindIdentityPersonalToken(context.Context, *BindIdentityPersonalTokenRequest) (*BindIdentityPersonalTokenResponse, error) {
return nil, status.Error(codes.Unimplemented, "method BindIdentityPersonalToken not implemented")
}
func (UnimplementedAccountServiceServer) mustEmbedUnimplementedAccountServiceServer() {}
func (UnimplementedAccountServiceServer) testEmbeddedByValue() {}
@@ -720,6 +746,24 @@ func _AccountService_RevokePersonalToken_Handler(srv interface{}, ctx context.Co
return interceptor(ctx, in, info, handler)
}
func _AccountService_BindIdentityPersonalToken_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
in := new(BindIdentityPersonalTokenRequest)
if err := dec(in); err != nil {
return nil, err
}
if interceptor == nil {
return srv.(AccountServiceServer).BindIdentityPersonalToken(ctx, in)
}
info := &grpc.UnaryServerInfo{
Server: srv,
FullMethod: AccountService_BindIdentityPersonalToken_FullMethodName,
}
handler := func(ctx context.Context, req interface{}) (interface{}, error) {
return srv.(AccountServiceServer).BindIdentityPersonalToken(ctx, req.(*BindIdentityPersonalTokenRequest))
}
return interceptor(ctx, in, info, handler)
}
// AccountService_ServiceDesc is the grpc.ServiceDesc for AccountService service.
// It's only intended for direct use with grpc.RegisterService,
// and not to be introspected or modified (even as a copy)
@@ -803,6 +847,10 @@ var AccountService_ServiceDesc = grpc.ServiceDesc{
MethodName: "RevokePersonalToken",
Handler: _AccountService_RevokePersonalToken_Handler,
},
{
MethodName: "BindIdentityPersonalToken",
Handler: _AccountService_BindIdentityPersonalToken_Handler,
},
},
Streams: []grpc.StreamDesc{},
Metadata: "state/v1/account.proto",
+228 -112
View File
@@ -2163,6 +2163,110 @@ func (x *EnqueueCallTaskRequest) GetCallKey() string {
return ""
}
type HandoffTaskRequest struct {
state protoimpl.MessageState `protogen:"open.v1"`
TaskId int64 `protobuf:"varint,1,opt,name=task_id,json=taskId,proto3" json:"task_id,omitempty"`
Workflow string `protobuf:"bytes,2,opt,name=workflow,proto3" json:"workflow,omitempty"`
Detail string `protobuf:"bytes,3,opt,name=detail,proto3" json:"detail,omitempty"`
unknownFields protoimpl.UnknownFields
sizeCache protoimpl.SizeCache
}
func (x *HandoffTaskRequest) Reset() {
*x = HandoffTaskRequest{}
mi := &file_state_v1_task_proto_msgTypes[41]
ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
ms.StoreMessageInfo(mi)
}
func (x *HandoffTaskRequest) String() string {
return protoimpl.X.MessageStringOf(x)
}
func (*HandoffTaskRequest) ProtoMessage() {}
func (x *HandoffTaskRequest) ProtoReflect() protoreflect.Message {
mi := &file_state_v1_task_proto_msgTypes[41]
if x != nil {
ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
if ms.LoadMessageInfo() == nil {
ms.StoreMessageInfo(mi)
}
return ms
}
return mi.MessageOf(x)
}
// Deprecated: Use HandoffTaskRequest.ProtoReflect.Descriptor instead.
func (*HandoffTaskRequest) Descriptor() ([]byte, []int) {
return file_state_v1_task_proto_rawDescGZIP(), []int{41}
}
func (x *HandoffTaskRequest) GetTaskId() int64 {
if x != nil {
return x.TaskId
}
return 0
}
func (x *HandoffTaskRequest) GetWorkflow() string {
if x != nil {
return x.Workflow
}
return ""
}
func (x *HandoffTaskRequest) GetDetail() string {
if x != nil {
return x.Detail
}
return ""
}
type HandoffTaskResponse struct {
state protoimpl.MessageState `protogen:"open.v1"`
Task *Task `protobuf:"bytes,1,opt,name=task,proto3" json:"task,omitempty"`
unknownFields protoimpl.UnknownFields
sizeCache protoimpl.SizeCache
}
func (x *HandoffTaskResponse) Reset() {
*x = HandoffTaskResponse{}
mi := &file_state_v1_task_proto_msgTypes[42]
ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
ms.StoreMessageInfo(mi)
}
func (x *HandoffTaskResponse) String() string {
return protoimpl.X.MessageStringOf(x)
}
func (*HandoffTaskResponse) ProtoMessage() {}
func (x *HandoffTaskResponse) ProtoReflect() protoreflect.Message {
mi := &file_state_v1_task_proto_msgTypes[42]
if x != nil {
ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
if ms.LoadMessageInfo() == nil {
ms.StoreMessageInfo(mi)
}
return ms
}
return mi.MessageOf(x)
}
// Deprecated: Use HandoffTaskResponse.ProtoReflect.Descriptor instead.
func (*HandoffTaskResponse) Descriptor() ([]byte, []int) {
return file_state_v1_task_proto_rawDescGZIP(), []int{42}
}
func (x *HandoffTaskResponse) GetTask() *Task {
if x != nil {
return x.Task
}
return nil
}
type EnqueueCallTaskResponse struct {
state protoimpl.MessageState `protogen:"open.v1"`
Task *Task `protobuf:"bytes,1,opt,name=task,proto3" json:"task,omitempty"`
@@ -2173,7 +2277,7 @@ type EnqueueCallTaskResponse struct {
func (x *EnqueueCallTaskResponse) Reset() {
*x = EnqueueCallTaskResponse{}
mi := &file_state_v1_task_proto_msgTypes[41]
mi := &file_state_v1_task_proto_msgTypes[43]
ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
ms.StoreMessageInfo(mi)
}
@@ -2185,7 +2289,7 @@ func (x *EnqueueCallTaskResponse) String() string {
func (*EnqueueCallTaskResponse) ProtoMessage() {}
func (x *EnqueueCallTaskResponse) ProtoReflect() protoreflect.Message {
mi := &file_state_v1_task_proto_msgTypes[41]
mi := &file_state_v1_task_proto_msgTypes[43]
if x != nil {
ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
if ms.LoadMessageInfo() == nil {
@@ -2198,7 +2302,7 @@ func (x *EnqueueCallTaskResponse) ProtoReflect() protoreflect.Message {
// Deprecated: Use EnqueueCallTaskResponse.ProtoReflect.Descriptor instead.
func (*EnqueueCallTaskResponse) Descriptor() ([]byte, []int) {
return file_state_v1_task_proto_rawDescGZIP(), []int{41}
return file_state_v1_task_proto_rawDescGZIP(), []int{43}
}
func (x *EnqueueCallTaskResponse) GetTask() *Task {
@@ -2225,7 +2329,7 @@ type WorkflowCallStatusRequest struct {
func (x *WorkflowCallStatusRequest) Reset() {
*x = WorkflowCallStatusRequest{}
mi := &file_state_v1_task_proto_msgTypes[42]
mi := &file_state_v1_task_proto_msgTypes[44]
ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
ms.StoreMessageInfo(mi)
}
@@ -2237,7 +2341,7 @@ func (x *WorkflowCallStatusRequest) String() string {
func (*WorkflowCallStatusRequest) ProtoMessage() {}
func (x *WorkflowCallStatusRequest) ProtoReflect() protoreflect.Message {
mi := &file_state_v1_task_proto_msgTypes[42]
mi := &file_state_v1_task_proto_msgTypes[44]
if x != nil {
ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
if ms.LoadMessageInfo() == nil {
@@ -2250,7 +2354,7 @@ func (x *WorkflowCallStatusRequest) ProtoReflect() protoreflect.Message {
// Deprecated: Use WorkflowCallStatusRequest.ProtoReflect.Descriptor instead.
func (*WorkflowCallStatusRequest) Descriptor() ([]byte, []int) {
return file_state_v1_task_proto_rawDescGZIP(), []int{42}
return file_state_v1_task_proto_rawDescGZIP(), []int{44}
}
func (x *WorkflowCallStatusRequest) GetCallerTaskId() int64 {
@@ -2283,7 +2387,7 @@ type WorkflowCallStatusResponse struct {
func (x *WorkflowCallStatusResponse) Reset() {
*x = WorkflowCallStatusResponse{}
mi := &file_state_v1_task_proto_msgTypes[43]
mi := &file_state_v1_task_proto_msgTypes[45]
ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
ms.StoreMessageInfo(mi)
}
@@ -2295,7 +2399,7 @@ func (x *WorkflowCallStatusResponse) String() string {
func (*WorkflowCallStatusResponse) ProtoMessage() {}
func (x *WorkflowCallStatusResponse) ProtoReflect() protoreflect.Message {
mi := &file_state_v1_task_proto_msgTypes[43]
mi := &file_state_v1_task_proto_msgTypes[45]
if x != nil {
ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
if ms.LoadMessageInfo() == nil {
@@ -2308,7 +2412,7 @@ func (x *WorkflowCallStatusResponse) ProtoReflect() protoreflect.Message {
// Deprecated: Use WorkflowCallStatusResponse.ProtoReflect.Descriptor instead.
func (*WorkflowCallStatusResponse) Descriptor() ([]byte, []int) {
return file_state_v1_task_proto_rawDescGZIP(), []int{43}
return file_state_v1_task_proto_rawDescGZIP(), []int{45}
}
func (x *WorkflowCallStatusResponse) GetStatus() string {
@@ -2341,7 +2445,7 @@ type TaskModelsRequest struct {
func (x *TaskModelsRequest) Reset() {
*x = TaskModelsRequest{}
mi := &file_state_v1_task_proto_msgTypes[44]
mi := &file_state_v1_task_proto_msgTypes[46]
ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
ms.StoreMessageInfo(mi)
}
@@ -2353,7 +2457,7 @@ func (x *TaskModelsRequest) String() string {
func (*TaskModelsRequest) ProtoMessage() {}
func (x *TaskModelsRequest) ProtoReflect() protoreflect.Message {
mi := &file_state_v1_task_proto_msgTypes[44]
mi := &file_state_v1_task_proto_msgTypes[46]
if x != nil {
ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
if ms.LoadMessageInfo() == nil {
@@ -2366,7 +2470,7 @@ func (x *TaskModelsRequest) ProtoReflect() protoreflect.Message {
// Deprecated: Use TaskModelsRequest.ProtoReflect.Descriptor instead.
func (*TaskModelsRequest) Descriptor() ([]byte, []int) {
return file_state_v1_task_proto_rawDescGZIP(), []int{44}
return file_state_v1_task_proto_rawDescGZIP(), []int{46}
}
func (x *TaskModelsRequest) GetTaskId() int64 {
@@ -2385,7 +2489,7 @@ type TaskInstructionsRequest struct {
func (x *TaskInstructionsRequest) Reset() {
*x = TaskInstructionsRequest{}
mi := &file_state_v1_task_proto_msgTypes[45]
mi := &file_state_v1_task_proto_msgTypes[47]
ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
ms.StoreMessageInfo(mi)
}
@@ -2397,7 +2501,7 @@ func (x *TaskInstructionsRequest) String() string {
func (*TaskInstructionsRequest) ProtoMessage() {}
func (x *TaskInstructionsRequest) ProtoReflect() protoreflect.Message {
mi := &file_state_v1_task_proto_msgTypes[45]
mi := &file_state_v1_task_proto_msgTypes[47]
if x != nil {
ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
if ms.LoadMessageInfo() == nil {
@@ -2410,7 +2514,7 @@ func (x *TaskInstructionsRequest) ProtoReflect() protoreflect.Message {
// Deprecated: Use TaskInstructionsRequest.ProtoReflect.Descriptor instead.
func (*TaskInstructionsRequest) Descriptor() ([]byte, []int) {
return file_state_v1_task_proto_rawDescGZIP(), []int{45}
return file_state_v1_task_proto_rawDescGZIP(), []int{47}
}
func (x *TaskInstructionsRequest) GetTaskId() int64 {
@@ -2430,7 +2534,7 @@ type TaskInstructionsResponse struct {
func (x *TaskInstructionsResponse) Reset() {
*x = TaskInstructionsResponse{}
mi := &file_state_v1_task_proto_msgTypes[46]
mi := &file_state_v1_task_proto_msgTypes[48]
ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
ms.StoreMessageInfo(mi)
}
@@ -2442,7 +2546,7 @@ func (x *TaskInstructionsResponse) String() string {
func (*TaskInstructionsResponse) ProtoMessage() {}
func (x *TaskInstructionsResponse) ProtoReflect() protoreflect.Message {
mi := &file_state_v1_task_proto_msgTypes[46]
mi := &file_state_v1_task_proto_msgTypes[48]
if x != nil {
ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
if ms.LoadMessageInfo() == nil {
@@ -2455,7 +2559,7 @@ func (x *TaskInstructionsResponse) ProtoReflect() protoreflect.Message {
// Deprecated: Use TaskInstructionsResponse.ProtoReflect.Descriptor instead.
func (*TaskInstructionsResponse) Descriptor() ([]byte, []int) {
return file_state_v1_task_proto_rawDescGZIP(), []int{46}
return file_state_v1_task_proto_rawDescGZIP(), []int{48}
}
func (x *TaskInstructionsResponse) GetInstance() string {
@@ -2483,7 +2587,7 @@ type TaskModelsResponse struct {
func (x *TaskModelsResponse) Reset() {
*x = TaskModelsResponse{}
mi := &file_state_v1_task_proto_msgTypes[47]
mi := &file_state_v1_task_proto_msgTypes[49]
ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
ms.StoreMessageInfo(mi)
}
@@ -2495,7 +2599,7 @@ func (x *TaskModelsResponse) String() string {
func (*TaskModelsResponse) ProtoMessage() {}
func (x *TaskModelsResponse) ProtoReflect() protoreflect.Message {
mi := &file_state_v1_task_proto_msgTypes[47]
mi := &file_state_v1_task_proto_msgTypes[49]
if x != nil {
ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
if ms.LoadMessageInfo() == nil {
@@ -2508,7 +2612,7 @@ func (x *TaskModelsResponse) ProtoReflect() protoreflect.Message {
// Deprecated: Use TaskModelsResponse.ProtoReflect.Descriptor instead.
func (*TaskModelsResponse) Descriptor() ([]byte, []int) {
return file_state_v1_task_proto_rawDescGZIP(), []int{47}
return file_state_v1_task_proto_rawDescGZIP(), []int{49}
}
func (x *TaskModelsResponse) GetOrgId() string {
@@ -2550,7 +2654,7 @@ type ModelProvider struct {
func (x *ModelProvider) Reset() {
*x = ModelProvider{}
mi := &file_state_v1_task_proto_msgTypes[48]
mi := &file_state_v1_task_proto_msgTypes[50]
ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
ms.StoreMessageInfo(mi)
}
@@ -2562,7 +2666,7 @@ func (x *ModelProvider) String() string {
func (*ModelProvider) ProtoMessage() {}
func (x *ModelProvider) ProtoReflect() protoreflect.Message {
mi := &file_state_v1_task_proto_msgTypes[48]
mi := &file_state_v1_task_proto_msgTypes[50]
if x != nil {
ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
if ms.LoadMessageInfo() == nil {
@@ -2575,7 +2679,7 @@ func (x *ModelProvider) ProtoReflect() protoreflect.Message {
// Deprecated: Use ModelProvider.ProtoReflect.Descriptor instead.
func (*ModelProvider) Descriptor() ([]byte, []int) {
return file_state_v1_task_proto_rawDescGZIP(), []int{48}
return file_state_v1_task_proto_rawDescGZIP(), []int{50}
}
func (x *ModelProvider) GetClass() string {
@@ -2746,7 +2850,13 @@ const file_state_v1_task_proto_rawDesc = "" +
"\bworkflow\x18\x02 \x01(\tR\bworkflow\x12\x1f\n" +
"\vinputs_json\x18\x03 \x01(\tR\n" +
"inputsJson\x12\x19\n" +
"\bcall_key\x18\x04 \x01(\tR\acallKey\"Y\n" +
"\bcall_key\x18\x04 \x01(\tR\acallKey\"a\n" +
"\x12HandoffTaskRequest\x12\x17\n" +
"\atask_id\x18\x01 \x01(\x03R\x06taskId\x12\x1a\n" +
"\bworkflow\x18\x02 \x01(\tR\bworkflow\x12\x16\n" +
"\x06detail\x18\x03 \x01(\tR\x06detail\"9\n" +
"\x13HandoffTaskResponse\x12\"\n" +
"\x04task\x18\x01 \x01(\v2\x0e.state.v1.TaskR\x04task\"Y\n" +
"\x17EnqueueCallTaskResponse\x12\"\n" +
"\x04task\x18\x01 \x01(\v2\x0e.state.v1.TaskR\x04task\x12\x1a\n" +
"\baccepted\x18\x02 \x01(\bR\baccepted\"c\n" +
@@ -2780,7 +2890,7 @@ const file_state_v1_task_proto_rawDesc = "" +
"\bbase_url\x18\x02 \x01(\tR\abaseUrl\x12\x1e\n" +
"\vapi_key_env\x18\x03 \x01(\tR\tapiKeyEnv\x12$\n" +
"\x0eapi_key_engine\x18\x04 \x01(\tR\fapiKeyEngine\x12\x1e\n" +
"\vapi_key_key\x18\x05 \x01(\tR\tapiKeyKey2\xe6\x0e\n" +
"\vapi_key_key\x18\x05 \x01(\tR\tapiKeyKey2\xb2\x0f\n" +
"\vTaskService\x12M\n" +
"\fEnqueueIssue\x12\x1d.state.v1.EnqueueIssueRequest\x1a\x1e.state.v1.EnqueueIssueResponse\x12V\n" +
"\x0fEnqueueChatTask\x12 .state.v1.EnqueueChatTaskRequest\x1a!.state.v1.EnqueueChatTaskResponse\x12b\n" +
@@ -2806,7 +2916,8 @@ const file_state_v1_task_proto_rawDesc = "" +
"\aOpenPRs\x12\x18.state.v1.OpenPRsRequest\x1a\x19.state.v1.OpenPRsResponse\x12_\n" +
"\x12ClearTerminalTasks\x12#.state.v1.ClearTerminalTasksRequest\x1a$.state.v1.ClearTerminalTasksResponse\x128\n" +
"\x05Tasks\x12\x16.state.v1.TasksRequest\x1a\x17.state.v1.TasksResponse\x12V\n" +
"\x0fEnqueueCallTask\x12 .state.v1.EnqueueCallTaskRequest\x1a!.state.v1.EnqueueCallTaskResponse\x12_\n" +
"\x0fEnqueueCallTask\x12 .state.v1.EnqueueCallTaskRequest\x1a!.state.v1.EnqueueCallTaskResponse\x12J\n" +
"\vHandoffTask\x12\x1c.state.v1.HandoffTaskRequest\x1a\x1d.state.v1.HandoffTaskResponse\x12_\n" +
"\x12WorkflowCallStatus\x12#.state.v1.WorkflowCallStatusRequest\x1a$.state.v1.WorkflowCallStatusResponseB\xa3\x01\n" +
"\fcom.state.v1B\tTaskProtoP\x01ZGgithub.com/samcharles93/archie-core/internal/contracts/state/v1;statev1\xa2\x02\x03SXX\xaa\x02\bState.V1\xca\x02\bState\\V1\xe2\x02\x14State\\V1\\GPBMetadata\xea\x02\tState::V1b\x06proto3"
@@ -2822,7 +2933,7 @@ func file_state_v1_task_proto_rawDescGZIP() []byte {
return file_state_v1_task_proto_rawDescData
}
var file_state_v1_task_proto_msgTypes = make([]protoimpl.MessageInfo, 51)
var file_state_v1_task_proto_msgTypes = make([]protoimpl.MessageInfo, 53)
var file_state_v1_task_proto_goTypes = []any{
(*EnqueueIssueRequest)(nil), // 0: state.v1.EnqueueIssueRequest
(*EnqueueIssueResponse)(nil), // 1: state.v1.EnqueueIssueResponse
@@ -2865,91 +2976,96 @@ var file_state_v1_task_proto_goTypes = []any{
(*TasksRequest)(nil), // 38: state.v1.TasksRequest
(*TasksResponse)(nil), // 39: state.v1.TasksResponse
(*EnqueueCallTaskRequest)(nil), // 40: state.v1.EnqueueCallTaskRequest
(*EnqueueCallTaskResponse)(nil), // 41: state.v1.EnqueueCallTaskResponse
(*WorkflowCallStatusRequest)(nil), // 42: state.v1.WorkflowCallStatusRequest
(*WorkflowCallStatusResponse)(nil), // 43: state.v1.WorkflowCallStatusResponse
(*TaskModelsRequest)(nil), // 44: state.v1.TaskModelsRequest
(*TaskInstructionsRequest)(nil), // 45: state.v1.TaskInstructionsRequest
(*TaskInstructionsResponse)(nil), // 46: state.v1.TaskInstructionsResponse
(*TaskModelsResponse)(nil), // 47: state.v1.TaskModelsResponse
(*ModelProvider)(nil), // 48: state.v1.ModelProvider
nil, // 49: state.v1.TaskModelsResponse.AliasesEntry
nil, // 50: state.v1.TaskModelsResponse.ProvidersEntry
(*Task)(nil), // 51: state.v1.Task
(*Event)(nil), // 52: state.v1.Event
(*timestamppb.Timestamp)(nil), // 53: google.protobuf.Timestamp
(*HandoffTaskRequest)(nil), // 41: state.v1.HandoffTaskRequest
(*HandoffTaskResponse)(nil), // 42: state.v1.HandoffTaskResponse
(*EnqueueCallTaskResponse)(nil), // 43: state.v1.EnqueueCallTaskResponse
(*WorkflowCallStatusRequest)(nil), // 44: state.v1.WorkflowCallStatusRequest
(*WorkflowCallStatusResponse)(nil), // 45: state.v1.WorkflowCallStatusResponse
(*TaskModelsRequest)(nil), // 46: state.v1.TaskModelsRequest
(*TaskInstructionsRequest)(nil), // 47: state.v1.TaskInstructionsRequest
(*TaskInstructionsResponse)(nil), // 48: state.v1.TaskInstructionsResponse
(*TaskModelsResponse)(nil), // 49: state.v1.TaskModelsResponse
(*ModelProvider)(nil), // 50: state.v1.ModelProvider
nil, // 51: state.v1.TaskModelsResponse.AliasesEntry
nil, // 52: state.v1.TaskModelsResponse.ProvidersEntry
(*Task)(nil), // 53: state.v1.Task
(*Event)(nil), // 54: state.v1.Event
(*timestamppb.Timestamp)(nil), // 55: google.protobuf.Timestamp
}
var file_state_v1_task_proto_depIdxs = []int32{
51, // 0: state.v1.ActiveTasksByOriginResponse.tasks:type_name -> state.v1.Task
51, // 1: state.v1.UnpostedOutcomesResponse.tasks:type_name -> state.v1.Task
51, // 2: state.v1.EnqueueChatTaskResponse.task:type_name -> state.v1.Task
51, // 3: state.v1.ClaimNextResponse.task:type_name -> state.v1.Task
51, // 4: state.v1.ClaimByIssueResponse.task:type_name -> state.v1.Task
51, // 5: state.v1.UpdateRequest.task:type_name -> state.v1.Task
52, // 6: state.v1.ArchiveTaskRequest.audit:type_name -> state.v1.Event
51, // 7: state.v1.TaskByIssueResponse.task:type_name -> state.v1.Task
51, // 8: state.v1.OpenTaskByPRResponse.task:type_name -> state.v1.Task
51, // 9: state.v1.TaskByIDResponse.task:type_name -> state.v1.Task
51, // 10: state.v1.OpenPRsResponse.tasks:type_name -> state.v1.Task
53, // 11: state.v1.TasksRequest.after_updated_at:type_name -> google.protobuf.Timestamp
51, // 12: state.v1.TasksResponse.tasks:type_name -> state.v1.Task
51, // 13: state.v1.EnqueueCallTaskResponse.task:type_name -> state.v1.Task
49, // 14: state.v1.TaskModelsResponse.aliases:type_name -> state.v1.TaskModelsResponse.AliasesEntry
50, // 15: state.v1.TaskModelsResponse.providers:type_name -> state.v1.TaskModelsResponse.ProvidersEntry
48, // 16: state.v1.TaskModelsResponse.ProvidersEntry.value:type_name -> state.v1.ModelProvider
0, // 17: state.v1.TaskService.EnqueueIssue:input_type -> state.v1.EnqueueIssueRequest
8, // 18: state.v1.TaskService.EnqueueChatTask:input_type -> state.v1.EnqueueChatTaskRequest
2, // 19: state.v1.TaskService.ActiveTasksByOrigin:input_type -> state.v1.ActiveTasksByOriginRequest
4, // 20: state.v1.TaskService.UnpostedOutcomes:input_type -> state.v1.UnpostedOutcomesRequest
6, // 21: state.v1.TaskService.MarkOutcomePosted:input_type -> state.v1.MarkOutcomePostedRequest
10, // 22: state.v1.TaskService.ClaimNext:input_type -> state.v1.ClaimNextRequest
12, // 23: state.v1.TaskService.ClaimByIssue:input_type -> state.v1.ClaimByIssueRequest
14, // 24: state.v1.TaskService.Transition:input_type -> state.v1.TransitionRequest
16, // 25: state.v1.TaskService.Update:input_type -> state.v1.UpdateRequest
44, // 26: state.v1.TaskService.TaskModels:input_type -> state.v1.TaskModelsRequest
45, // 27: state.v1.TaskService.TaskInstructions:input_type -> state.v1.TaskInstructionsRequest
18, // 28: state.v1.TaskService.Requeue:input_type -> state.v1.RequeueRequest
20, // 29: state.v1.TaskService.ParkTask:input_type -> state.v1.ParkTaskRequest
22, // 30: state.v1.TaskService.RecoverStale:input_type -> state.v1.RecoverStaleRequest
24, // 31: state.v1.TaskService.ArchiveTask:input_type -> state.v1.ArchiveTaskRequest
26, // 32: state.v1.TaskService.RetryTask:input_type -> state.v1.RetryTaskRequest
28, // 33: state.v1.TaskService.TaskByIssue:input_type -> state.v1.TaskByIssueRequest
30, // 34: state.v1.TaskService.OpenTaskByPR:input_type -> state.v1.OpenTaskByPRRequest
32, // 35: state.v1.TaskService.TaskByID:input_type -> state.v1.TaskByIDRequest
34, // 36: state.v1.TaskService.OpenPRs:input_type -> state.v1.OpenPRsRequest
36, // 37: state.v1.TaskService.ClearTerminalTasks:input_type -> state.v1.ClearTerminalTasksRequest
38, // 38: state.v1.TaskService.Tasks:input_type -> state.v1.TasksRequest
40, // 39: state.v1.TaskService.EnqueueCallTask:input_type -> state.v1.EnqueueCallTaskRequest
42, // 40: state.v1.TaskService.WorkflowCallStatus:input_type -> state.v1.WorkflowCallStatusRequest
1, // 41: state.v1.TaskService.EnqueueIssue:output_type -> state.v1.EnqueueIssueResponse
9, // 42: state.v1.TaskService.EnqueueChatTask:output_type -> state.v1.EnqueueChatTaskResponse
3, // 43: state.v1.TaskService.ActiveTasksByOrigin:output_type -> state.v1.ActiveTasksByOriginResponse
5, // 44: state.v1.TaskService.UnpostedOutcomes:output_type -> state.v1.UnpostedOutcomesResponse
7, // 45: state.v1.TaskService.MarkOutcomePosted:output_type -> state.v1.MarkOutcomePostedResponse
11, // 46: state.v1.TaskService.ClaimNext:output_type -> state.v1.ClaimNextResponse
13, // 47: state.v1.TaskService.ClaimByIssue:output_type -> state.v1.ClaimByIssueResponse
15, // 48: state.v1.TaskService.Transition:output_type -> state.v1.TransitionResponse
17, // 49: state.v1.TaskService.Update:output_type -> state.v1.UpdateResponse
47, // 50: state.v1.TaskService.TaskModels:output_type -> state.v1.TaskModelsResponse
46, // 51: state.v1.TaskService.TaskInstructions:output_type -> state.v1.TaskInstructionsResponse
19, // 52: state.v1.TaskService.Requeue:output_type -> state.v1.RequeueResponse
21, // 53: state.v1.TaskService.ParkTask:output_type -> state.v1.ParkTaskResponse
23, // 54: state.v1.TaskService.RecoverStale:output_type -> state.v1.RecoverStaleResponse
25, // 55: state.v1.TaskService.ArchiveTask:output_type -> state.v1.ArchiveTaskResponse
27, // 56: state.v1.TaskService.RetryTask:output_type -> state.v1.RetryTaskResponse
29, // 57: state.v1.TaskService.TaskByIssue:output_type -> state.v1.TaskByIssueResponse
31, // 58: state.v1.TaskService.OpenTaskByPR:output_type -> state.v1.OpenTaskByPRResponse
33, // 59: state.v1.TaskService.TaskByID:output_type -> state.v1.TaskByIDResponse
35, // 60: state.v1.TaskService.OpenPRs:output_type -> state.v1.OpenPRsResponse
37, // 61: state.v1.TaskService.ClearTerminalTasks:output_type -> state.v1.ClearTerminalTasksResponse
39, // 62: state.v1.TaskService.Tasks:output_type -> state.v1.TasksResponse
41, // 63: state.v1.TaskService.EnqueueCallTask:output_type -> state.v1.EnqueueCallTaskResponse
43, // 64: state.v1.TaskService.WorkflowCallStatus:output_type -> state.v1.WorkflowCallStatusResponse
41, // [41:65] is the sub-list for method output_type
17, // [17:41] is the sub-list for method input_type
17, // [17:17] is the sub-list for extension type_name
17, // [17:17] is the sub-list for extension extendee
0, // [0:17] is the sub-list for field type_name
53, // 0: state.v1.ActiveTasksByOriginResponse.tasks:type_name -> state.v1.Task
53, // 1: state.v1.UnpostedOutcomesResponse.tasks:type_name -> state.v1.Task
53, // 2: state.v1.EnqueueChatTaskResponse.task:type_name -> state.v1.Task
53, // 3: state.v1.ClaimNextResponse.task:type_name -> state.v1.Task
53, // 4: state.v1.ClaimByIssueResponse.task:type_name -> state.v1.Task
53, // 5: state.v1.UpdateRequest.task:type_name -> state.v1.Task
54, // 6: state.v1.ArchiveTaskRequest.audit:type_name -> state.v1.Event
53, // 7: state.v1.TaskByIssueResponse.task:type_name -> state.v1.Task
53, // 8: state.v1.OpenTaskByPRResponse.task:type_name -> state.v1.Task
53, // 9: state.v1.TaskByIDResponse.task:type_name -> state.v1.Task
53, // 10: state.v1.OpenPRsResponse.tasks:type_name -> state.v1.Task
55, // 11: state.v1.TasksRequest.after_updated_at:type_name -> google.protobuf.Timestamp
53, // 12: state.v1.TasksResponse.tasks:type_name -> state.v1.Task
53, // 13: state.v1.HandoffTaskResponse.task:type_name -> state.v1.Task
53, // 14: state.v1.EnqueueCallTaskResponse.task:type_name -> state.v1.Task
51, // 15: state.v1.TaskModelsResponse.aliases:type_name -> state.v1.TaskModelsResponse.AliasesEntry
52, // 16: state.v1.TaskModelsResponse.providers:type_name -> state.v1.TaskModelsResponse.ProvidersEntry
50, // 17: state.v1.TaskModelsResponse.ProvidersEntry.value:type_name -> state.v1.ModelProvider
0, // 18: state.v1.TaskService.EnqueueIssue:input_type -> state.v1.EnqueueIssueRequest
8, // 19: state.v1.TaskService.EnqueueChatTask:input_type -> state.v1.EnqueueChatTaskRequest
2, // 20: state.v1.TaskService.ActiveTasksByOrigin:input_type -> state.v1.ActiveTasksByOriginRequest
4, // 21: state.v1.TaskService.UnpostedOutcomes:input_type -> state.v1.UnpostedOutcomesRequest
6, // 22: state.v1.TaskService.MarkOutcomePosted:input_type -> state.v1.MarkOutcomePostedRequest
10, // 23: state.v1.TaskService.ClaimNext:input_type -> state.v1.ClaimNextRequest
12, // 24: state.v1.TaskService.ClaimByIssue:input_type -> state.v1.ClaimByIssueRequest
14, // 25: state.v1.TaskService.Transition:input_type -> state.v1.TransitionRequest
16, // 26: state.v1.TaskService.Update:input_type -> state.v1.UpdateRequest
46, // 27: state.v1.TaskService.TaskModels:input_type -> state.v1.TaskModelsRequest
47, // 28: state.v1.TaskService.TaskInstructions:input_type -> state.v1.TaskInstructionsRequest
18, // 29: state.v1.TaskService.Requeue:input_type -> state.v1.RequeueRequest
20, // 30: state.v1.TaskService.ParkTask:input_type -> state.v1.ParkTaskRequest
22, // 31: state.v1.TaskService.RecoverStale:input_type -> state.v1.RecoverStaleRequest
24, // 32: state.v1.TaskService.ArchiveTask:input_type -> state.v1.ArchiveTaskRequest
26, // 33: state.v1.TaskService.RetryTask:input_type -> state.v1.RetryTaskRequest
28, // 34: state.v1.TaskService.TaskByIssue:input_type -> state.v1.TaskByIssueRequest
30, // 35: state.v1.TaskService.OpenTaskByPR:input_type -> state.v1.OpenTaskByPRRequest
32, // 36: state.v1.TaskService.TaskByID:input_type -> state.v1.TaskByIDRequest
34, // 37: state.v1.TaskService.OpenPRs:input_type -> state.v1.OpenPRsRequest
36, // 38: state.v1.TaskService.ClearTerminalTasks:input_type -> state.v1.ClearTerminalTasksRequest
38, // 39: state.v1.TaskService.Tasks:input_type -> state.v1.TasksRequest
40, // 40: state.v1.TaskService.EnqueueCallTask:input_type -> state.v1.EnqueueCallTaskRequest
41, // 41: state.v1.TaskService.HandoffTask:input_type -> state.v1.HandoffTaskRequest
44, // 42: state.v1.TaskService.WorkflowCallStatus:input_type -> state.v1.WorkflowCallStatusRequest
1, // 43: state.v1.TaskService.EnqueueIssue:output_type -> state.v1.EnqueueIssueResponse
9, // 44: state.v1.TaskService.EnqueueChatTask:output_type -> state.v1.EnqueueChatTaskResponse
3, // 45: state.v1.TaskService.ActiveTasksByOrigin:output_type -> state.v1.ActiveTasksByOriginResponse
5, // 46: state.v1.TaskService.UnpostedOutcomes:output_type -> state.v1.UnpostedOutcomesResponse
7, // 47: state.v1.TaskService.MarkOutcomePosted:output_type -> state.v1.MarkOutcomePostedResponse
11, // 48: state.v1.TaskService.ClaimNext:output_type -> state.v1.ClaimNextResponse
13, // 49: state.v1.TaskService.ClaimByIssue:output_type -> state.v1.ClaimByIssueResponse
15, // 50: state.v1.TaskService.Transition:output_type -> state.v1.TransitionResponse
17, // 51: state.v1.TaskService.Update:output_type -> state.v1.UpdateResponse
49, // 52: state.v1.TaskService.TaskModels:output_type -> state.v1.TaskModelsResponse
48, // 53: state.v1.TaskService.TaskInstructions:output_type -> state.v1.TaskInstructionsResponse
19, // 54: state.v1.TaskService.Requeue:output_type -> state.v1.RequeueResponse
21, // 55: state.v1.TaskService.ParkTask:output_type -> state.v1.ParkTaskResponse
23, // 56: state.v1.TaskService.RecoverStale:output_type -> state.v1.RecoverStaleResponse
25, // 57: state.v1.TaskService.ArchiveTask:output_type -> state.v1.ArchiveTaskResponse
27, // 58: state.v1.TaskService.RetryTask:output_type -> state.v1.RetryTaskResponse
29, // 59: state.v1.TaskService.TaskByIssue:output_type -> state.v1.TaskByIssueResponse
31, // 60: state.v1.TaskService.OpenTaskByPR:output_type -> state.v1.OpenTaskByPRResponse
33, // 61: state.v1.TaskService.TaskByID:output_type -> state.v1.TaskByIDResponse
35, // 62: state.v1.TaskService.OpenPRs:output_type -> state.v1.OpenPRsResponse
37, // 63: state.v1.TaskService.ClearTerminalTasks:output_type -> state.v1.ClearTerminalTasksResponse
39, // 64: state.v1.TaskService.Tasks:output_type -> state.v1.TasksResponse
43, // 65: state.v1.TaskService.EnqueueCallTask:output_type -> state.v1.EnqueueCallTaskResponse
42, // 66: state.v1.TaskService.HandoffTask:output_type -> state.v1.HandoffTaskResponse
45, // 67: state.v1.TaskService.WorkflowCallStatus:output_type -> state.v1.WorkflowCallStatusResponse
43, // [43:68] is the sub-list for method output_type
18, // [18:43] is the sub-list for method input_type
18, // [18:18] is the sub-list for extension type_name
18, // [18:18] is the sub-list for extension extendee
0, // [0:18] is the sub-list for field type_name
}
func init() { file_state_v1_task_proto_init() }
@@ -2964,7 +3080,7 @@ func file_state_v1_task_proto_init() {
GoPackagePath: reflect.TypeOf(x{}).PkgPath(),
RawDescriptor: unsafe.Slice(unsafe.StringData(file_state_v1_task_proto_rawDesc), len(file_state_v1_task_proto_rawDesc)),
NumEnums: 0,
NumMessages: 51,
NumMessages: 53,
NumExtensions: 0,
NumServices: 1,
},
@@ -42,6 +42,7 @@ const (
TaskService_ClearTerminalTasks_FullMethodName = "/state.v1.TaskService/ClearTerminalTasks"
TaskService_Tasks_FullMethodName = "/state.v1.TaskService/Tasks"
TaskService_EnqueueCallTask_FullMethodName = "/state.v1.TaskService/EnqueueCallTask"
TaskService_HandoffTask_FullMethodName = "/state.v1.TaskService/HandoffTask"
TaskService_WorkflowCallStatus_FullMethodName = "/state.v1.TaskService/WorkflowCallStatus"
)
@@ -84,6 +85,9 @@ type TaskServiceClient interface {
// (docs/prds/workflow-calls.md). CallStatus re-checks the parent-child
// relation server-side, so the grant only names the caller.
EnqueueCallTask(ctx context.Context, in *EnqueueCallTaskRequest, opts ...grpc.CallOption) (*EnqueueCallTaskResponse, error)
// HandoffTask completes a running run and starts its linked successor on
// the same work item under another workflow.
HandoffTask(ctx context.Context, in *HandoffTaskRequest, opts ...grpc.CallOption) (*HandoffTaskResponse, error)
WorkflowCallStatus(ctx context.Context, in *WorkflowCallStatusRequest, opts ...grpc.CallOption) (*WorkflowCallStatusResponse, error)
}
@@ -325,6 +329,16 @@ func (c *taskServiceClient) EnqueueCallTask(ctx context.Context, in *EnqueueCall
return out, nil
}
func (c *taskServiceClient) HandoffTask(ctx context.Context, in *HandoffTaskRequest, opts ...grpc.CallOption) (*HandoffTaskResponse, error) {
cOpts := append([]grpc.CallOption{grpc.StaticMethod()}, opts...)
out := new(HandoffTaskResponse)
err := c.cc.Invoke(ctx, TaskService_HandoffTask_FullMethodName, in, out, cOpts...)
if err != nil {
return nil, err
}
return out, nil
}
func (c *taskServiceClient) WorkflowCallStatus(ctx context.Context, in *WorkflowCallStatusRequest, opts ...grpc.CallOption) (*WorkflowCallStatusResponse, error) {
cOpts := append([]grpc.CallOption{grpc.StaticMethod()}, opts...)
out := new(WorkflowCallStatusResponse)
@@ -374,6 +388,9 @@ type TaskServiceServer interface {
// (docs/prds/workflow-calls.md). CallStatus re-checks the parent-child
// relation server-side, so the grant only names the caller.
EnqueueCallTask(context.Context, *EnqueueCallTaskRequest) (*EnqueueCallTaskResponse, error)
// HandoffTask completes a running run and starts its linked successor on
// the same work item under another workflow.
HandoffTask(context.Context, *HandoffTaskRequest) (*HandoffTaskResponse, error)
WorkflowCallStatus(context.Context, *WorkflowCallStatusRequest) (*WorkflowCallStatusResponse, error)
mustEmbedUnimplementedTaskServiceServer()
}
@@ -454,6 +471,9 @@ func (UnimplementedTaskServiceServer) Tasks(context.Context, *TasksRequest) (*Ta
func (UnimplementedTaskServiceServer) EnqueueCallTask(context.Context, *EnqueueCallTaskRequest) (*EnqueueCallTaskResponse, error) {
return nil, status.Error(codes.Unimplemented, "method EnqueueCallTask not implemented")
}
func (UnimplementedTaskServiceServer) HandoffTask(context.Context, *HandoffTaskRequest) (*HandoffTaskResponse, error) {
return nil, status.Error(codes.Unimplemented, "method HandoffTask not implemented")
}
func (UnimplementedTaskServiceServer) WorkflowCallStatus(context.Context, *WorkflowCallStatusRequest) (*WorkflowCallStatusResponse, error) {
return nil, status.Error(codes.Unimplemented, "method WorkflowCallStatus not implemented")
}
@@ -892,6 +912,24 @@ func _TaskService_EnqueueCallTask_Handler(srv interface{}, ctx context.Context,
return interceptor(ctx, in, info, handler)
}
func _TaskService_HandoffTask_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
in := new(HandoffTaskRequest)
if err := dec(in); err != nil {
return nil, err
}
if interceptor == nil {
return srv.(TaskServiceServer).HandoffTask(ctx, in)
}
info := &grpc.UnaryServerInfo{
Server: srv,
FullMethod: TaskService_HandoffTask_FullMethodName,
}
handler := func(ctx context.Context, req interface{}) (interface{}, error) {
return srv.(TaskServiceServer).HandoffTask(ctx, req.(*HandoffTaskRequest))
}
return interceptor(ctx, in, info, handler)
}
func _TaskService_WorkflowCallStatus_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
in := new(WorkflowCallStatusRequest)
if err := dec(in); err != nil {
@@ -1009,6 +1047,10 @@ var TaskService_ServiceDesc = grpc.ServiceDesc{
MethodName: "EnqueueCallTask",
Handler: _TaskService_EnqueueCallTask_Handler,
},
{
MethodName: "HandoffTask",
Handler: _TaskService_HandoffTask_Handler,
},
{
MethodName: "WorkflowCallStatus",
Handler: _TaskService_WorkflowCallStatus_Handler,
+25 -5
View File
@@ -229,9 +229,13 @@ type Task struct {
// recorded at the park site; empty when the park has no known fix.
ParkBlocker string `protobuf:"bytes,48,opt,name=park_blocker,json=parkBlocker,proto3" json:"park_blocker,omitempty"`
// plan_document is the structured plan (JSON) plan was rendered from.
PlanDocument string `protobuf:"bytes,50,opt,name=plan_document,json=planDocument,proto3" json:"plan_document,omitempty"`
unknownFields protoimpl.UnknownFields
sizeCache protoimpl.SizeCache
PlanDocument string `protobuf:"bytes,50,opt,name=plan_document,json=planDocument,proto3" json:"plan_document,omitempty"`
// handoff_from_task_id is the run that handed off to this one; 0 when none.
HandoffFromTaskId int64 `protobuf:"varint,51,opt,name=handoff_from_task_id,json=handoffFromTaskId,proto3" json:"handoff_from_task_id,omitempty"`
// handoff_to_task_id is the run this one handed off to; 0 when none.
HandoffToTaskId int64 `protobuf:"varint,52,opt,name=handoff_to_task_id,json=handoffToTaskId,proto3" json:"handoff_to_task_id,omitempty"`
unknownFields protoimpl.UnknownFields
sizeCache protoimpl.SizeCache
}
func (x *Task) Reset() {
@@ -608,6 +612,20 @@ func (x *Task) GetPlanDocument() string {
return ""
}
func (x *Task) GetHandoffFromTaskId() int64 {
if x != nil {
return x.HandoffFromTaskId
}
return 0
}
func (x *Task) GetHandoffToTaskId() int64 {
if x != nil {
return x.HandoffToTaskId
}
return 0
}
// Event mirrors internal/events.Event. Data is carried as a JSON object
// string (matching the store's own on-disk representation) rather than
// google.protobuf.Struct, to stay a direct mirror of the existing column.
@@ -784,7 +802,7 @@ const file_state_v1_types_proto_rawDesc = "" +
"\n" +
"created_at\x18\x06 \x01(\v2\x1a.google.protobuf.TimestampR\tcreatedAt\x129\n" +
"\n" +
"updated_at\x18\a \x01(\v2\x1a.google.protobuf.TimestampR\tupdatedAt\"\x91\r\n" +
"updated_at\x18\a \x01(\v2\x1a.google.protobuf.TimestampR\tupdatedAt\"\xef\r\n" +
"\x04Task\x12\x0e\n" +
"\x02id\x18\x01 \x01(\x03R\x02id\x12\x14\n" +
"\x05owner\x18\x02 \x01(\tR\x05owner\x12\x12\n" +
@@ -849,7 +867,9 @@ const file_state_v1_types_proto_rawDesc = "" +
"\x11binding_signature\x18. \x01(\tR\x10bindingSignature\x12!\n" +
"\fbinding_addr\x18/ \x01(\tR\vbindingAddr\x12!\n" +
"\fpark_blocker\x180 \x01(\tR\vparkBlocker\x12#\n" +
"\rplan_document\x182 \x01(\tR\fplanDocumentJ\x04\b1\x102R\tplan_view\"\xf8\x02\n" +
"\rplan_document\x182 \x01(\tR\fplanDocument\x12/\n" +
"\x14handoff_from_task_id\x183 \x01(\x03R\x11handoffFromTaskId\x12+\n" +
"\x12handoff_to_task_id\x184 \x01(\x03R\x0fhandoffToTaskIdJ\x04\b1\x102R\tplan_view\"\xf8\x02\n" +
"\x05Event\x12\x0e\n" +
"\x02id\x18\x01 \x01(\x03R\x02id\x12*\n" +
"\x02at\x18\x02 \x01(\v2\x1a.google.protobuf.TimestampR\x02at\x12\x12\n" +
+3 -1
View File
@@ -1753,7 +1753,9 @@ func (d *Daemon) cleanupTerminalTaskWorktree(ctx context.Context, task *workflow
d.Log.WarnContext(ctx, "terminal worktree cleanup task lookup failed", "task", task.ID, "err", err)
return
}
if latest == nil {
// A run that handed off shares its work item's worktree with its
// successor, which owns it from here.
if latest == nil || latest.HandoffToTaskID != 0 {
return
}
// Settled statuses. Keep the worktree while a PR exists.
+59
View File
@@ -0,0 +1,59 @@
package access
import "github.com/samcharles93/archie-core/internal/domain/org"
// CanaryGrantID is the ID the canary's scoped grant is known by in the policy
// store and in denial records. A stored policy must not reuse it: the store's
// seeding and the reset command overwrite shipped IDs, and this one is
// installed by the operator from CanaryGrantPolicy's text.
const CanaryGrantID = "org-canary-grant"
// CanaryGrantForbidID is the matching forbid's ID: the grant only scopes the
// canary if everything else it could reach is forbidden, because the identity
// may also carry a role (API-created service accounts are assigned developer)
// whose shipped policy permits broadly within the org. A forbid wins within
// the level, so the pair scopes the identity regardless of its role.
const CanaryGrantForbidID = "org-canary-grant-forbid"
// canaryGrantActions is the exact action surface the canary needs, as one
// Cedar condition shared by the grant's permit and its forbid.
func canaryGrantActions() string {
return `(
(action == Archie::Action::"read" && ["dashboard", "task"].contains(resource.kind)) ||
(action == Archie::Action::"read" && resource == Archie::Object::"update-settings") ||
(action == Archie::Action::"update" && resource == Archie::Object::"update") ||
(action == Archie::Action::"manage_identities" && resource == Archie::Object::"update-settings")
)`
}
func canaryPrincipal(identityID string) string {
return `principal == Archie::Identity::"` + identityID + `"`
}
// CanaryGrantPolicy is the grant a CI canary carries instead of instance
// admin: exactly one named identity may read the dashboard's signal surface
// (chat update, alerts, logs, tasks), read and write the update-settings
// resource, and trigger an update install or deferral. Install it together
// with CanaryGrantForbidPolicy, which forbids the same identity from
// everything else. Every clause is guarded by the resource belonging to the
// principal's own org, which the engine enforces structurally anyway.
func CanaryGrantPolicy(orgID org.OrgID, identityID string) Policy {
text := `permit(principal, action, resource) when {
resource.org == principal.org &&
` + canaryPrincipal(identityID) + ` &&
` + canaryGrantActions() + `
};`
return Policy{ID: CanaryGrantID, Level: LevelOrg, OrgID: orgID, Text: text}
}
// CanaryGrantForbidPolicy forbids the canary identity from everything outside
// CanaryGrantPolicy's surface, so a role assigned to the identity (service
// accounts created through the API carry developer) cannot widen it.
func CanaryGrantForbidPolicy(orgID org.OrgID, identityID string) Policy {
text := `forbid(principal, action, resource) when {
resource.org == principal.org &&
` + canaryPrincipal(identityID) + ` &&
!` + canaryGrantActions() + `
};`
return Policy{ID: CanaryGrantForbidID, Level: LevelOrg, OrgID: orgID, Text: text}
}
+54
View File
@@ -0,0 +1,54 @@
package access
import (
"strings"
"testing"
"github.com/samcharles93/archie-core/internal/domain/org"
)
// TestCanaryGrantPolicy pins the shape of the canary grant: it names exactly
// the one identity it is built for, scoped to the org it serves, so a token
// carrying it cannot act as anyone else or reach another org.
func TestCanaryGrantPolicy(t *testing.T) {
p := CanaryGrantPolicy(org.DefaultOrgID, "canary-svc")
if err := p.Validate(); err != nil {
t.Fatalf("Validate: %v", err)
}
if p.ID != CanaryGrantID || p.Level != LevelOrg || p.OrgID != org.DefaultOrgID {
t.Fatalf("policy = %+v, want ID %q at org level in %q", p, CanaryGrantID, org.DefaultOrgID)
}
if !strings.Contains(p.Text, `principal == Archie::Identity::"canary-svc"`) {
t.Fatalf("text does not name the canary identity:\n%s", p.Text)
}
other := CanaryGrantPolicy(org.DefaultOrgID, "another")
if strings.Contains(other.Text, "canary-svc") || !strings.Contains(other.Text, `Archie::Identity::"another"`) {
t.Fatalf("grant for another identity must name that identity only:\n%s", other.Text)
}
}
// TestCanaryGrantForbidPolicy pins the forbid that keeps a role assigned to
// the canary identity from widening the grant: it names the same identity and
// negates exactly the permit's action surface.
func TestCanaryGrantForbidPolicy(t *testing.T) {
permit := CanaryGrantPolicy(org.DefaultOrgID, "canary-svc")
forbid := CanaryGrantForbidPolicy(org.DefaultOrgID, "canary-svc")
if err := forbid.Validate(); err != nil {
t.Fatalf("Validate: %v", err)
}
if forbid.ID != CanaryGrantForbidID || forbid.Level != LevelOrg || forbid.OrgID != org.DefaultOrgID {
t.Fatalf("forbid = %+v, want ID %q at org level in %q", forbid, CanaryGrantForbidID, org.DefaultOrgID)
}
if !strings.HasPrefix(forbid.Text, "forbid(") {
t.Fatalf("forbid text must be a forbid:\n%s", forbid.Text)
}
if !strings.Contains(forbid.Text, `principal == Archie::Identity::"canary-svc"`) {
t.Fatalf("forbid does not name the canary identity:\n%s", forbid.Text)
}
// The forbid must negate the permit's surface, not the permit itself.
permitBody := strings.TrimSuffix(strings.SplitN(permit.Text, "when {", 2)[1], "};")
forbidBody := strings.TrimSuffix(strings.SplitN(forbid.Text, "when {", 2)[1], "};")
if !strings.Contains(forbidBody, "!"+canaryGrantActions()) {
t.Fatalf("forbid does not negate the grant's action surface:\npermit:%s\nforbid:%s", permitBody, forbidBody)
}
}
+13 -1
View File
@@ -13,6 +13,10 @@ import (
// registry's activity surface and events.
const ActionToolCalled = "tool.called"
// ActionSummary is the Action.Type recorded for the pass's final text: the
// report a definition pass produces, kept inspectable in the activity trail.
const ActionSummary = "pass.summary"
// defaultDefinitionMaxSteps bounds the model/tool loop of one definition pass.
const defaultDefinitionMaxSteps = 10
@@ -84,7 +88,7 @@ func (e *DefinitionEngine) Pass(ctx context.Context, in PassInput) (PassResult,
return PassResult{}, err
}
actions := make([]Action, 0, len(res.ToolCalls))
actions := make([]Action, 0, len(res.ToolCalls)+1)
for _, call := range res.ToolCalls {
actions = append(actions, Action{
At: e.host.Clock.Now(),
@@ -93,6 +97,14 @@ func (e *DefinitionEngine) Pass(ctx context.Context, in PassInput) (PassResult,
Reason: in.Reason,
})
}
if summary := strings.TrimSpace(res.Text); summary != "" {
actions = append(actions, Action{
At: e.host.Clock.Now(),
Type: ActionSummary,
Detail: summary,
Reason: in.Reason,
})
}
return PassResult{Actions: actions}, nil
}
@@ -139,3 +139,31 @@ func TestInstructionOnlyRegistrarViewFiltered(t *testing.T) {
t.Error("registrar view has no clock; time is always bound")
}
}
// TestDefinitionPassRecordsSummary pins that a pass's final text is recorded
// as a pass.summary action, so a definition's report stays inspectable in the
// activity trail instead of being dropped when the model returns.
func TestDefinitionPassRecordsSummary(t *testing.T) {
llm := &fakeLLM{}
registry := NewRegistry(fullHost(llm))
engine := NewDefinitionEngine(instructionOnlyDefinition())
if err := registry.Register(engine); err != nil {
t.Fatal(err)
}
res, err := engine.Pass(t.Context(), PassInput{Reason: "test"})
if err != nil {
t.Fatalf("pass: %v", err)
}
var summary *Action
for i, a := range res.Actions {
if a.Type == ActionSummary {
summary = &res.Actions[i]
}
}
if summary == nil {
t.Fatalf("no %s action recorded; actions: %+v", ActionSummary, res.Actions)
}
if summary.Detail != "done" {
t.Errorf("summary detail = %q, want the model's final text %q", summary.Detail, "done")
}
}
+1 -1
View File
@@ -76,7 +76,7 @@ func mustKey(encoded string) ed25519.PublicKey {
func DefaultCatalogue() Source {
return Source{
Reference: "git.catlow.cloud/sam/archipelago/catalogue",
Key: mustKey("YL4nvn07uq/5mXZOt9cEmLYu/kB6C2+MLTnbHyoX4Pw="),
Key: mustKey("PLouShOzvRqy1QbYaJplaM41/ulgKFN8ZQ7g1rI02Fc="),
}
}
+87
View File
@@ -0,0 +1,87 @@
package storecontract
import (
"context"
"time"
"github.com/samcharles93/archie-core/internal/domain/binding"
"github.com/samcharles93/archie-core/internal/domain/workflow/task"
)
// BindingStore persists bindings and their draft -> pending_approval ->
// armed states.
type BindingStore interface {
InsertBinding(ctx context.Context, b binding.Binding) (string, error)
GetBinding(ctx context.Context, id string) (*binding.Binding, error)
ListBindings(ctx context.Context) ([]binding.Binding, error)
UpdateBinding(ctx context.Context, b binding.Binding) error
DeleteBinding(ctx context.Context, id string) error
ApproveBinding(ctx context.Context, id string) error
PauseBinding(ctx context.Context, id string) error
ResumeBinding(ctx context.Context, id string) error
}
// DispatchLedger reads the binding dispatch ledger.
type DispatchLedger interface {
// ListDispatches returns the newest ledger rows matching filter.
ListDispatches(ctx context.Context, filter DispatchFilter) ([]Dispatch, error)
}
// Dispatch is one binding/capture ledger row: the task it started, or the
// reason it started none.
type Dispatch struct {
BindingID string `json:"binding_id"`
BindingVersion int64 `json:"binding_version"`
CaptureID string `json:"capture_id"`
TaskID int64 `json:"task_id,omitempty"`
Reason string `json:"reason,omitempty"`
DispatchedAt time.Time `json:"dispatched_at"`
}
// DispatchFilter narrows ListDispatches; empty fields match any row.
type DispatchFilter struct {
BindingID string
CaptureID string
TaskID int64
Limit int
}
// BindingDispatcher looks up armed bindings, lists undispatched captures and
// records dispatches.
type BindingDispatcher interface {
ArmedBindingsForSource(ctx context.Context, source string) ([]binding.Binding, error)
RecordDispatch(
ctx context.Context,
bindingID string,
bindingVersion int64,
captureID string,
// taskID addresses the SQLite-owned task tables, which are not
// migrating: it stays an integer on purpose. A dispatch that starts
// a task is recorded by EnqueueBindingTask instead.
taskID int64,
// reason is empty for a dispatch and names the terminal outcome of
// an evaluation that started no task.
reason string,
) error
ListUndispatchedCaptures(ctx context.Context, bindings []string, limit int) ([]CapturedEvent, error)
}
// BindingTaskCreator is the single-method consumer-facing surface for
// enqueueing a task triggered by a binding. Splitting it off TaskLifecycle
// keeps the lifecycle surface narrow (8 methods, the interfacebloat limit)
// and keeps the binding-specific shape on the binding interfaces.
type BindingTaskCreator interface {
// EnqueueBindingTask claims the binding's dispatch of captureID in the
// ledger, enqueues the task and records its provenance as one unit: an
// error leaves none of them, and a delivery already dispatched for the
// binding returns ErrAlreadyDispatched.
EnqueueBindingTask(ctx context.Context, owner, repo, title, body, wf, identity, bindingID string, bindingVersion int, captureID string, inputs map[string]any) (*task.Task, error)
}
// PlaybookDispatcher records a playbook action dispatch before its side
// effect runs, returning ErrAlreadyDispatched for a duplicate.
// DeletePlaybookDispatches deletes one playbook's records.
type PlaybookDispatcher interface {
RecordPlaybookDispatch(ctx context.Context, playbookID, playbookVersion, eventID, actionID string) error
DeletePlaybookDispatches(ctx context.Context, playbookID string) error
}
+73
View File
@@ -0,0 +1,73 @@
package storecontract
import (
"context"
"time"
"github.com/samcharles93/archie-core/internal/domain/eventtype"
)
// CaptureStore persists inbound webhook captures.
type CaptureStore interface {
InsertCapture(ctx context.Context, c CapturedEvent, retention time.Duration, maxEvents int) (string, error)
ListCaptures(ctx context.Context, limit int) ([]CapturedEvent, error)
}
// CaptureRefusalStore counts the events network rules refused, per source.
type CaptureRefusalStore interface {
RecordCaptureRefusal(ctx context.Context, source, addr string, at time.Time) error
CaptureRefusals(ctx context.Context, since time.Time) ([]CaptureRefusals, error)
}
// CaptureRefusals is one source's refusals and accepted captures since a
// time, and the addresses it refused.
type CaptureRefusals struct {
Source string `json:"source"`
Refused int64 `json:"refused"`
Accepted int64 `json:"accepted"`
Addrs []string `json:"addrs"`
}
// EventTypeStore persists event types. Insert and Update refuse a type whose
// rule overlaps another on the same source with eventtype.ErrOverlap, and a
// malformed one with eventtype.ErrInvalid. Update rewrites the name and rule;
// the schema is the one the type was created with.
type EventTypeStore interface {
InsertEventType(ctx context.Context, t eventtype.EventType) (string, error)
UpdateEventType(ctx context.Context, t eventtype.EventType) error
DeleteEventType(ctx context.Context, id string) error
ListEventTypes(ctx context.Context) ([]eventtype.EventType, error)
}
// CapturedEvent is one unbound inbound webhook capture: no workflow binding,
// no forge/task association -- just what arrived, from where, and when.
type CapturedEvent struct {
ID string `json:"id"`
ReceivedAt time.Time `json:"received_at"`
Source string `json:"source"`
RemoteAddr string `json:"remote_addr"`
ContentType string `json:"content_type"`
// Headers and Body are redacted (webhookguard.RedactPayload) before
// they ever reach InsertCapture; the store persists what it is given.
Headers string `json:"headers"`
Body string `json:"body"`
Authenticated bool `json:"authenticated"`
// EventType is the ID of the event type the capture was identified as
// when it arrived, or empty when it matched none. An unidentified capture
// is never dispatched.
EventType string `json:"event_type"`
// Unsigned marks an event that arrived on an approved unsigned source.
// It dispatches without Authenticated and is flagged wherever it shows.
Unsigned bool `json:"unsigned"`
// Delivery names the sender's delivery; a retry of one delivery shares
// it, so a binding dispatches each delivery once. Intake sets it from the
// source's delivery header; the store fills the capture ID when empty.
Delivery string `json:"delivery"`
// OrgID is honoured only for a schedule-fired capture, whose source is
// reserved and so names no org; every other capture takes its source's.
OrgID string `json:"org_id,omitempty"`
}
// Dispatchable reports whether a binding may start a task from this event:
// a valid signature, or a source approved to take unsigned events.
func (c CapturedEvent) Dispatchable() bool { return c.Authenticated || c.Unsigned }
@@ -0,0 +1,25 @@
package storecontract
import (
"context"
"time"
"github.com/samcharles93/archie-core/internal/domain/harnesssecret"
)
// HarnessSecretStore stores one org's OAuth token set per credential binding
// service, encrypted at rest.
type HarnessSecretStore interface {
GetHarnessSecret(ctx context.Context, org, service string) (harnesssecret.Secret, error)
PutHarnessSecret(ctx context.Context, s harnesssecret.Secret) error
}
// RunCredentialStore keeps run credentials by digest: the SHA-256 of the
// token a task's container presents. A lookup of an unknown or expired
// digest reports ErrRunCredentialUnknown.
type RunCredentialStore interface {
PutRunCredential(ctx context.Context, digest [32]byte, taskID int64, expires time.Time) error
DeleteRunCredential(ctx context.Context, digest [32]byte) error
RunCredentialTask(ctx context.Context, digest [32]byte, now time.Time) (int64, error)
DeleteExpiredRunCredentials(ctx context.Context, now time.Time) error
}
+90
View File
@@ -0,0 +1,90 @@
package storecontract
import (
"errors"
)
// Sentinel errors, part of the wire contract: staterpc's
// mapError/unmapError match these by (code, canonical message) across the
// gRPC boundary, so their message strings are load-bearing and must never
// change.
var (
// ErrStaleTransition is returned when a task transition's expected
// from status does not match the task's current status.
ErrStaleTransition = errors.New("store: stale transition: task status does not match expected from status")
// ErrIllegalTransition is returned when the transition table does not allow
// the status pair.
ErrIllegalTransition = errors.New("store: illegal transition: status pair is not in the transition table")
// ErrInvalidStep is returned when a step-execution request names a kind the
// store does not record or a step whose identifying fields do not resolve.
// Distinct from the transition sentinels, which cover a row's state.
ErrInvalidStep = errors.New("store: invalid step execution request")
// ErrResumeIncomplete is returned when a retry resumes after a stage whose
// last run did not complete, so there are no results to start from.
ErrResumeIncomplete = errors.New("store: the step before the resume point did not complete in its last run")
// ErrBindingNotFound is returned when a binding ID does not exist.
ErrBindingNotFound = errors.New("store: binding not found")
// ErrBindingOverlap is returned when a binding's matcher overlaps an
// existing binding for the same source.
ErrBindingOverlap = errors.New("store: binding overlaps existing binding for source")
// ErrBindingTransition is returned when a binding state transition is
// rejected by the draft -> pending_approval -> armed machine.
ErrBindingTransition = errors.New("store: binding state transition rejected")
// ErrAlreadyDispatched is returned when a dispatch ledger already holds the
// key.
ErrAlreadyDispatched = errors.New("store: binding already dispatched for capture")
// ErrCallNotYours is returned when a workflow.call status read names a
// task that is not a callee of the calling task, or the caller itself
// does not exist: a caller reads only the tasks it started
ErrCallNotYours = errors.New("store: task is not a callee of the calling task")
// ErrCallCallerNotRunning is returned when a workflow.call enqueue names
// a caller that is not running: a stale retry, or a bug in the caller.
ErrCallCallerNotRunning = errors.New("store: workflow.call caller is not running")
// ErrCallDepthExceeded is returned when a workflow.call enqueue would
// pass workflow.MaxCallDepth. The engine refuses the same call first;
// the store re-checks because it owns the table.
ErrCallDepthExceeded = errors.New("store: workflow.call would pass the depth limit")
// ErrMappingNotFound is returned when a mapping ID does not exist.
ErrMappingNotFound = errors.New("store: mapping not found")
// ErrEventTypeNotFound is returned when an event type ID does not exist.
ErrEventTypeNotFound = errors.New("store: event type not found")
// ErrSourceNotFound is returned when a source path does not exist.
ErrSourceNotFound = errors.New("store: source not found")
// ErrSourceInUse is returned when deleting a source an armed binding
// fires on.
ErrSourceInUse = errors.New("store: source has an armed binding")
// ErrSourcePathTaken is returned when a new source's path is in use.
ErrSourcePathTaken = errors.New("store: source path already taken")
// ErrSourceSigningStale is returned when a signing write's expected
// from state does not match the stored one.
ErrSourceSigningStale = errors.New("store: source signing does not match expected state")
// ErrHarnessSecretNotFound is returned when no OAuth token set is
// stored for an org/service pair -- the setup terminal has not
// captured one yet.
ErrHarnessSecretNotFound = errors.New("store: harness secret not found")
// ErrRereviewCapReached is returned when a review gate re-review would
// round past prreview.MaxRereviewRounds. The task stays waiting and its
// gate row is unchanged. The dashboard answers it 409, like every other
// conflict.
ErrRereviewCapReached = errors.New("store: re-review cap reached")
// ErrResourceNotFound is returned when a control-plane resource kind has
// no stored document. It is in-process only (not on the gRPC wire): the
// control plane runs in the daemon against the local store.
ErrResourceNotFound = errors.New("resource not found")
// ErrResourceVersionConflict is returned when a resource write carries an
// expected version that does not match the stored revision.
ErrResourceVersionConflict = errors.New("resource version conflict")
)
// ErrUnavailable means the store could not serve a request now: it is
// restarting or unreachable. It is not paired with a canonical message the way
// the sentinels above are -- staterpc derives it from the Unavailable code
// alone, whose text varies by transport -- and it is the one store error a
// caller with budget may answer by retrying rather than by failing its own
// work. The wrapping keeps the gRPC status code readable through the chain.
var ErrUnavailable = errors.New("store: state store unavailable")
// ErrRunCredentialUnknown is a run credential that was never issued, was
// revoked, or has expired.
var ErrRunCredentialUnknown = errors.New("run credential unknown or expired")
+26
View File
@@ -0,0 +1,26 @@
package storecontract
import (
"context"
"github.com/samcharles93/archie-core/internal/domain/mapping"
)
// MappingStore persists payload field mappings. Deliberately separate from
// TaskStore and CaptureStore for the same reason those are split: the
// dashboard's mapping editor should only acquire the mapping surface, not the
// full task or capture APIs. See
type MappingStore interface {
InsertMapping(ctx context.Context, m mapping.Mapping) (string, error)
GetMapping(ctx context.Context, id string) (*mapping.Mapping, error)
ListMappings(ctx context.Context) ([]mapping.Mapping, error)
UpdateMapping(ctx context.Context, m mapping.Mapping) error
DeleteMapping(ctx context.Context, id string) error
}
// MappingMatchRecorder counts the events each mapping resolved, which the
// store reports as Mapping.MatchCount and LastMatchedAt. Recording the same
// (mapping, capture) twice counts once.
type MappingMatchRecorder interface {
RecordMappingMatch(ctx context.Context, mappingID, captureID string) error
}
+50
View File
@@ -0,0 +1,50 @@
package storecontract
import (
"time"
)
// Resource is one control-plane resource document: the stored value for an
// org's kind, its revision, and the audit record written with the last write.
// Resource and ResourceWrite live here so the UI-side consumers can reference
// them without linking the PostgreSQL implementation.
type Resource struct {
OrgID string
Kind string
Value []byte
Version int64
Actor string
Source string
RequestID string
ExpectedVersion int64
CurrentVersion int64
At time.Time
}
// DefaultOrgID is the org every resource written before orgs existed belongs
// to, and the org a request that names none acts in.
const DefaultOrgID = "org-sys"
// ResourceAuditKey is the sys_audit record key of one org's resource kind. The
// default org keeps the bare kind, so audit rows written before resources were
// per-org still resolve.
func ResourceAuditKey(orgID, kind string) string {
if orgID == DefaultOrgID {
return kind
}
return orgID + "/" + kind
}
// ResourceWrite is a control-plane resource write request. ExpectedVersion is
// the optimistic-concurrency guard: a write whose expectation does not match
// the stored revision is refused with ErrResourceVersionConflict.
type ResourceWrite struct {
OrgID string
Kind string
Value []byte
Actor string
Source string
RequestID string
ExpectedVersion int64
At time.Time
}
+23
View File
@@ -0,0 +1,23 @@
package storecontract
import (
"context"
"github.com/samcharles93/archie-core/internal/domain/source"
)
// SourceStore persists capture sources keyed by path. Secrets are encrypted
// at rest; signing transitions use an expected-from guard.
type SourceStore interface {
InsertSource(ctx context.Context, s source.Source) error
// GetSource returns (nil, nil) for an unknown path.
GetSource(ctx context.Context, path string) (*source.Source, error)
ListSources(ctx context.Context) ([]source.Source, error)
SetSourceSigning(ctx context.Context, path string, from, to source.Signing) error
SetSourceSecret(ctx context.Context, path, secret string) error
SetSourceName(ctx context.Context, path, name string) error
SetSourceDeliveryHeader(ctx context.Context, path, header string) error
// DeleteSource refuses with ErrSourceInUse while an armed binding fires
// on the source.
DeleteSource(ctx context.Context, path string) error
}
+87
View File
@@ -0,0 +1,87 @@
package storecontract
import (
"context"
"time"
"github.com/samcharles93/archie-core/internal/domain/health"
)
// ConfigSnapshotStore holds the published configuration view.
type ConfigSnapshotStore interface {
PutConfigSnapshot(ctx context.Context, snapshot ConfigSnapshot) error
ConfigSnapshot(ctx context.Context) (ConfigSnapshot, bool, error)
}
// ChannelStatusStore holds channel runtime state published by the process
// hosting the channels.
type ChannelStatusStore interface {
// PutChannelStatus records the reporting process's whole set: a channel it no
// longer reports is removed rather than left behind, because a stale "running"
// row is a lie the dashboard would show indefinitely.
PutChannelStatus(ctx context.Context, channels []ChannelStatus) error
ChannelStatus(ctx context.Context) ([]ChannelStatus, error)
}
// ApplyStatusStore holds which version of each control-plane resource each
// process is running. Same split as ConfigSnapshotStore and for the same
// reason: the writers are the processes that apply a resource, the reader is
// the UI, and a task-scoped credential must reach neither.
type ApplyStatusStore interface {
PutApplyStatus(ctx context.Context, status ApplyStatus) error
ListApplyStatus(ctx context.Context) ([]ApplyStatus, error)
}
// PresenceStore holds one record per running service instance. Writers are
// the services themselves; readers are the UI and archied status.
type PresenceStore interface {
PutPresence(ctx context.Context, presence Presence) error
ListPresence(ctx context.Context) ([]Presence, error)
}
// ConfigSnapshot is the published, secret-free configuration view. Schema
// names Document's shape.
type ConfigSnapshot struct {
Schema string
Document []byte
PublishedAt time.Time
}
// ChannelStatus is one channel's runtime state as the process hosting it
// reports it. ReloadSupported is a capability declaration rather than a request:
// asking for a reload travels on its own surface, because a transient request and
// a durable state do not share a lifetime.
type ChannelStatus struct {
ID string
Name string
State string
Detail string
Configured bool
ReloadSupported bool
ObservedAt time.Time
}
// ApplyStatus is one process's report on one resource kind. On error,
// AppliedVersion stays at the last applied version. ReportedAt is
// re-stamped periodically.
type ApplyStatus struct {
Process string `json:"process"`
Kind string `json:"kind"`
AppliedVersion int64 `json:"applied_version"`
Error string `json:"error,omitempty"`
ReportedAt time.Time `json:"reported_at"`
}
// Presence is one running service instance. Ready and Detail summarise its
// readiness probes; Detail names the degraded ones.
type Presence struct {
Checks []health.Component `json:"checks"`
Service string `json:"service"`
InstanceID string `json:"instance_id"`
Version string `json:"version"`
InstallType string `json:"install_type"`
StartedAt time.Time `json:"started_at"`
ReportedAt time.Time `json:"reported_at"`
Ready bool `json:"ready"`
Detail string `json:"detail,omitempty"`
}
@@ -2,23 +2,6 @@
// without an implementation.
package storecontract
import (
"context"
"errors"
"io"
"time"
"github.com/samcharles93/archie-core/internal/domain/binding"
"github.com/samcharles93/archie-core/internal/domain/eventtype"
"github.com/samcharles93/archie-core/internal/domain/harnesssecret"
"github.com/samcharles93/archie-core/internal/domain/health"
"github.com/samcharles93/archie-core/internal/domain/mapping"
"github.com/samcharles93/archie-core/internal/domain/source"
"github.com/samcharles93/archie-core/internal/domain/workflow/task"
"github.com/samcharles93/archie-core/internal/events"
"github.com/samcharles93/archie-core/internal/logging"
)
// TaskStore is the full store surface the daemon needs.
// Consumers depend on TaskStore, never on a concrete type.
type TaskStore interface {
@@ -31,594 +14,3 @@ type TaskStore interface {
ReviewGateResponder
RemediationStarter
}
// TaskLifecycle manages the core task state machine and enqueuing.
// Binding-triggered task creation lives on BindingTaskCreator, not here,
// so the lifecycle surface stays narrow and non-binding consumers (forge
// poll, chat spawn, drain loop) do not acquire a binding-specific shape.
type TaskLifecycle interface {
EnqueueIssue(ctx context.Context, owner, repo string, number int, title, body, labels, identity string) (bool, error)
EnqueueChatTask(ctx context.Context, owner, repo, title, body, wf, identity, origin, draft string, inputs map[string]any) (*task.Task, error)
ClaimNext(ctx context.Context) (*task.Task, error)
ClaimByIssue(ctx context.Context, owner, repo string, number int) (*task.Task, error)
Transition(ctx context.Context, taskID int64, from, to, detail string) error
Update(ctx context.Context, t *task.Task) error
Requeue(ctx context.Context, taskID int64, fromStatus, workflow string) error
RecoverStale(ctx context.Context) (int64, error)
}
// TaskParker parks a task with a park class and the blocker the park site
// knows, if any. Unknown classes become needs_decision.
type TaskParker interface {
ParkTask(ctx context.Context, taskID int64, from, detail, class, blocker string) error
}
// StepRecorder records a running execution's steps.
type StepRecorder interface {
StartStep(ctx context.Context, s task.StepStart) (int64, events.Event, error)
FinishStep(ctx context.Context, s task.StepFinish) (events.Event, error)
}
// ExecutionCanceller cancels an execution and its current attempt's open
// steps in one transaction. It also cancels the callees the execution was
// waiting on, and returns their ids (nested callees included) so the caller
// can stop their runs too.
type ExecutionCanceller interface {
CancelExecution(ctx context.Context, taskID int64, reason, to string) ([]int64, error)
}
// StepReader lists one execution's recorded steps.
type StepReader interface {
ListSteps(ctx context.Context, executionID int64, attempt int) ([]task.StepExecution, error)
}
// HarnessSecretStore stores one org's OAuth token set per credential binding
// service, encrypted at rest.
type HarnessSecretStore interface {
GetHarnessSecret(ctx context.Context, org, service string) (harnesssecret.Secret, error)
PutHarnessSecret(ctx context.Context, s harnesssecret.Secret) error
}
// TaskArchiver removes one terminal task's local record with an optimistic
// status guard. It is separate from the already broad lifecycle contract so
// consumers that only run tasks do not acquire an operator-only capability.
type TaskArchiver interface {
OutcomePoster
ArchiveTask(ctx context.Context, taskID int64, fromStatus string, audit events.Event) (eventID int64, err error)
ClearTerminalTasks(ctx context.Context) (int64, error)
}
// TaskRetryer atomically requeues recoverable work and accounts for the new
// attempt so a partial write cannot evade the retry cap. retryMode is the
// operator's worktree choice for the next dispatch (taskstate.RetryMode),
// persisted on the row so the daemon reads it instead of inferring it. resume
// is where the attempt starts; a resume whose After stage did not complete in
// its last run is refused with ErrResumeIncomplete.
type TaskRetryer interface {
RetryTask(ctx context.Context, taskID int64, fromStatus, workflow, retryMode string, resume task.Resume, currentDefinition bool) error
}
// ReviewGateResponder records the operator's review-gate answer. A re-review
// increments rereview_rounds and requeues; one past the cap returns
// ErrRereviewCapReached and writes nothing.
type ReviewGateResponder interface {
RespondReviewGate(ctx context.Context, taskID int64, fromStatus, gate string, rereview bool, maxRounds int) error
}
// RemediationStarter queues a review-triggered remediation for an Archie-owned
// open-PR task: one guarded transition that carries the JSON-encoded review
// unit in the same write, so a claimed remediation always has its input. Runs
// for one task are serialised, so a distinct review that arrives while a
// remediation owns the task is queued behind it and remediated once that run
// finishes; a re-delivered unit is absorbed, not queued twice.
// UpdateReviewPayload merges late-arriving comments into the unclaimed unit
// for the same review, active or queued behind a running remediation.
type RemediationStarter interface {
BeginRemediation(ctx context.Context, taskID int64, payload string) error
UpdateReviewPayload(ctx context.Context, taskID int64, payload string) error
// SetReviewCursors persists the poll backstop's per-task review and
// comment high-water marks (pr-review-remediation.md decision 2). The
// pr_open guard is part of the contract: cursors only move while no
// remediation run owns the task.
SetReviewCursors(ctx context.Context, taskID, reviewCursor, commentCursor int64) error
}
// OutcomePoster tracks which finished chat tasks have reported back to their
// conversation.
type OutcomePoster interface {
// UnpostedOutcomes returns finished chat tasks whose outcome has not been
// posted to the conversation that created them.
UnpostedOutcomes(ctx context.Context) ([]*task.Task, error)
// MarkOutcomePosted records that the task's state at status was posted.
MarkOutcomePosted(ctx context.Context, taskID int64, status string) error
}
// TaskQueries groups read-only task accessors.
type TaskQueries interface {
TaskByIssue(ctx context.Context, owner, repo string, number int) (*task.Task, error)
OpenTaskByPR(ctx context.Context, owner, repo string, number int) (*task.Task, error)
TaskByID(ctx context.Context, taskID int64) (*task.Task, error)
OpenPRs(ctx context.Context) ([]task.Task, error)
// ActiveTasksByOrigin returns the queued and running tasks the
// conversation origin created, for its /stop.
ActiveTasksByOrigin(ctx context.Context, origin string) ([]*task.Task, error)
// Tasks returns the most recently updated tasks.
Tasks(ctx context.Context, limit int) ([]task.Task, error)
// TasksPage returns one page of tasks, most recently updated first.
TasksPage(ctx context.Context, page TaskPage) ([]task.Task, error)
StatusCounts(ctx context.Context) (map[string]int, error)
}
// TaskEvents groups observability and lifecycle methods.
type TaskEvents interface {
InsertEvent(ctx context.Context, e events.Event) (int64, error)
EventsSince(ctx context.Context, cursor string, limit int) ([]events.Event, error)
// TaskEvents returns a task's timeline, without agent transcripts.
TaskEvents(ctx context.Context, taskID int64) ([]events.Event, error)
// TaskTranscript returns one attempt's agent transcript entries in order.
TaskTranscript(ctx context.Context, taskID int64, attempt int) ([]events.Event, error)
WorkflowStats(ctx context.Context) ([]WorkflowStat, error)
StageStats(ctx context.Context) ([]StageStat, error)
TokensByDay(ctx context.Context, days int) ([]DayTokens, error)
Close() error
}
// CaptureStore persists inbound webhook captures.
type CaptureStore interface {
InsertCapture(ctx context.Context, c CapturedEvent, retention time.Duration, maxEvents int) (string, error)
ListCaptures(ctx context.Context, limit int) ([]CapturedEvent, error)
}
// CaptureRefusalStore counts the events network rules refused, per source.
type CaptureRefusalStore interface {
RecordCaptureRefusal(ctx context.Context, source, addr string, at time.Time) error
CaptureRefusals(ctx context.Context, since time.Time) ([]CaptureRefusals, error)
}
// CaptureRefusals is one source's refusals and accepted captures since a
// time, and the addresses it refused.
type CaptureRefusals struct {
Source string `json:"source"`
Refused int64 `json:"refused"`
Accepted int64 `json:"accepted"`
Addrs []string `json:"addrs"`
}
// ConfigSnapshotStore holds the published configuration view.
type ConfigSnapshotStore interface {
PutConfigSnapshot(ctx context.Context, snapshot ConfigSnapshot) error
ConfigSnapshot(ctx context.Context) (ConfigSnapshot, bool, error)
}
// ChannelStatusStore holds channel runtime state published by the process
// hosting the channels.
type ChannelStatusStore interface {
// PutChannelStatus records the reporting process's whole set: a channel it no
// longer reports is removed rather than left behind, because a stale "running"
// row is a lie the dashboard would show indefinitely.
PutChannelStatus(ctx context.Context, channels []ChannelStatus) error
ChannelStatus(ctx context.Context) ([]ChannelStatus, error)
}
// ApplyStatusStore holds which version of each control-plane resource each
// process is running. Same split as ConfigSnapshotStore and for the same
// reason: the writers are the processes that apply a resource, the reader is
// the UI, and a task-scoped credential must reach neither.
type ApplyStatusStore interface {
PutApplyStatus(ctx context.Context, status ApplyStatus) error
ListApplyStatus(ctx context.Context) ([]ApplyStatus, error)
}
// PresenceStore holds one record per running service instance. Writers are
// the services themselves; readers are the UI and archied status.
type PresenceStore interface {
PutPresence(ctx context.Context, presence Presence) error
ListPresence(ctx context.Context) ([]Presence, error)
}
// MappingStore persists payload field mappings. Deliberately separate from
// TaskStore and CaptureStore for the same reason those are split: the
// dashboard's mapping editor should only acquire the mapping surface, not the
// full task or capture APIs. See
type MappingStore interface {
InsertMapping(ctx context.Context, m mapping.Mapping) (string, error)
GetMapping(ctx context.Context, id string) (*mapping.Mapping, error)
ListMappings(ctx context.Context) ([]mapping.Mapping, error)
UpdateMapping(ctx context.Context, m mapping.Mapping) error
DeleteMapping(ctx context.Context, id string) error
}
// MappingMatchRecorder counts the events each mapping resolved, which the
// store reports as Mapping.MatchCount and LastMatchedAt. Recording the same
// (mapping, capture) twice counts once.
type MappingMatchRecorder interface {
RecordMappingMatch(ctx context.Context, mappingID, captureID string) error
}
// EventTypeStore persists event types. Insert and Update refuse a type whose
// rule overlaps another on the same source with eventtype.ErrOverlap, and a
// malformed one with eventtype.ErrInvalid. Update rewrites the name and rule;
// the schema is the one the type was created with.
type EventTypeStore interface {
InsertEventType(ctx context.Context, t eventtype.EventType) (string, error)
UpdateEventType(ctx context.Context, t eventtype.EventType) error
DeleteEventType(ctx context.Context, id string) error
ListEventTypes(ctx context.Context) ([]eventtype.EventType, error)
}
// BindingStore persists bindings and their draft -> pending_approval ->
// armed states.
type BindingStore interface {
InsertBinding(ctx context.Context, b binding.Binding) (string, error)
GetBinding(ctx context.Context, id string) (*binding.Binding, error)
ListBindings(ctx context.Context) ([]binding.Binding, error)
UpdateBinding(ctx context.Context, b binding.Binding) error
DeleteBinding(ctx context.Context, id string) error
ApproveBinding(ctx context.Context, id string) error
PauseBinding(ctx context.Context, id string) error
ResumeBinding(ctx context.Context, id string) error
}
// SourceStore persists capture sources keyed by path. Secrets are encrypted
// at rest; signing transitions use an expected-from guard.
type SourceStore interface {
InsertSource(ctx context.Context, s source.Source) error
// GetSource returns (nil, nil) for an unknown path.
GetSource(ctx context.Context, path string) (*source.Source, error)
ListSources(ctx context.Context) ([]source.Source, error)
SetSourceSigning(ctx context.Context, path string, from, to source.Signing) error
SetSourceSecret(ctx context.Context, path, secret string) error
SetSourceName(ctx context.Context, path, name string) error
SetSourceDeliveryHeader(ctx context.Context, path, header string) error
// DeleteSource refuses with ErrSourceInUse while an armed binding fires
// on the source.
DeleteSource(ctx context.Context, path string) error
}
// DispatchLedger reads the binding dispatch ledger.
type DispatchLedger interface {
// ListDispatches returns the newest ledger rows matching filter.
ListDispatches(ctx context.Context, filter DispatchFilter) ([]Dispatch, error)
}
// TaskPage selects one page of the task list. Empty Statuses matches every
// status. A zero After starts at the newest task; otherwise the page begins
// after the task it names, in (updated_at, id) descending order.
type TaskPage struct {
Statuses []string
// Workflow, when set, keeps only that workflow's tasks.
Workflow string
After TaskCursor
Limit int
}
// TaskCursor is the position of the last task a page returned.
type TaskCursor struct {
UpdatedAt time.Time
ID int64
}
// Dispatch is one binding/capture ledger row: the task it started, or the
// reason it started none.
type Dispatch struct {
BindingID string `json:"binding_id"`
BindingVersion int64 `json:"binding_version"`
CaptureID string `json:"capture_id"`
TaskID int64 `json:"task_id,omitempty"`
Reason string `json:"reason,omitempty"`
DispatchedAt time.Time `json:"dispatched_at"`
}
// DispatchFilter narrows ListDispatches; empty fields match any row.
type DispatchFilter struct {
BindingID string
CaptureID string
TaskID int64
Limit int
}
// BindingDispatcher looks up armed bindings, lists undispatched captures and
// records dispatches.
type BindingDispatcher interface {
ArmedBindingsForSource(ctx context.Context, source string) ([]binding.Binding, error)
RecordDispatch(
ctx context.Context,
bindingID string,
bindingVersion int64,
captureID string,
// taskID addresses the SQLite-owned task tables, which are not
// migrating: it stays an integer on purpose. A dispatch that starts
// a task is recorded by EnqueueBindingTask instead.
taskID int64,
// reason is empty for a dispatch and names the terminal outcome of
// an evaluation that started no task.
reason string,
) error
ListUndispatchedCaptures(ctx context.Context, bindings []string, limit int) ([]CapturedEvent, error)
}
// BindingTaskCreator is the single-method consumer-facing surface for
// enqueueing a task triggered by a binding. Splitting it off TaskLifecycle
// keeps the lifecycle surface narrow (8 methods, the interfacebloat limit)
// and keeps the binding-specific shape on the binding interfaces.
type BindingTaskCreator interface {
// EnqueueBindingTask claims the binding's dispatch of captureID in the
// ledger, enqueues the task and records its provenance as one unit: an
// error leaves none of them, and a delivery already dispatched for the
// binding returns ErrAlreadyDispatched.
EnqueueBindingTask(ctx context.Context, owner, repo, title, body, wf, identity, bindingID string, bindingVersion int, captureID string, inputs map[string]any) (*task.Task, error)
}
// WorkflowCaller starts a workflow.call callee and reads it back.
type WorkflowCaller = task.Caller
// PlaybookDispatcher records a playbook action dispatch before its side
// effect runs, returning ErrAlreadyDispatched for a duplicate.
// DeletePlaybookDispatches deletes one playbook's records.
type PlaybookDispatcher interface {
RecordPlaybookDispatch(ctx context.Context, playbookID, playbookVersion, eventID, actionID string) error
DeletePlaybookDispatches(ctx context.Context, playbookID string) error
}
// TaskLogStore reads one task attempt's log. Found=false means no log;
// ErrTaskLogsUnavailable means this process cannot read logs.
type TaskLogStore interface {
// TaskLog returns a page of one attempt's decoded log. attempt 0 selects
// the task's current attempt.
TaskLog(ctx context.Context, taskID int64, attempt int, q logging.Query) (logging.TaskLogPage, error)
// TaskLogContent writes one attempt's log verbatim to w, for a download.
// found is false, with a nil error, when the attempt has no log file.
TaskLogContent(ctx context.Context, taskID int64, attempt int, w io.Writer) (found bool, err error)
}
// CapturedEvent is one unbound inbound webhook capture: no workflow binding,
// no forge/task association -- just what arrived, from where, and when.
type CapturedEvent struct {
ID string `json:"id"`
ReceivedAt time.Time `json:"received_at"`
Source string `json:"source"`
RemoteAddr string `json:"remote_addr"`
ContentType string `json:"content_type"`
// Headers and Body are redacted (webhookguard.RedactPayload) before
// they ever reach InsertCapture; the store persists what it is given.
Headers string `json:"headers"`
Body string `json:"body"`
Authenticated bool `json:"authenticated"`
// EventType is the ID of the event type the capture was identified as
// when it arrived, or empty when it matched none. An unidentified capture
// is never dispatched.
EventType string `json:"event_type"`
// Unsigned marks an event that arrived on an approved unsigned source.
// It dispatches without Authenticated and is flagged wherever it shows.
Unsigned bool `json:"unsigned"`
// Delivery names the sender's delivery; a retry of one delivery shares
// it, so a binding dispatches each delivery once. Intake sets it from the
// source's delivery header; the store fills the capture ID when empty.
Delivery string `json:"delivery"`
// OrgID is honoured only for a schedule-fired capture, whose source is
// reserved and so names no org; every other capture takes its source's.
OrgID string `json:"org_id,omitempty"`
}
// Dispatchable reports whether a binding may start a task from this event:
// a valid signature, or a source approved to take unsigned events.
func (c CapturedEvent) Dispatchable() bool { return c.Authenticated || c.Unsigned }
// ConfigSnapshot is the published, secret-free configuration view. Schema
// names Document's shape.
type ConfigSnapshot struct {
Schema string
Document []byte
PublishedAt time.Time
}
// ChannelStatus is one channel's runtime state as the process hosting it
// reports it. ReloadSupported is a capability declaration rather than a request:
// asking for a reload travels on its own surface, because a transient request and
// a durable state do not share a lifetime.
type ChannelStatus struct {
ID string
Name string
State string
Detail string
Configured bool
ReloadSupported bool
ObservedAt time.Time
}
// ApplyStatus is one process's report on one resource kind. On error,
// AppliedVersion stays at the last applied version. ReportedAt is
// re-stamped periodically.
type ApplyStatus struct {
Process string `json:"process"`
Kind string `json:"kind"`
AppliedVersion int64 `json:"applied_version"`
Error string `json:"error,omitempty"`
ReportedAt time.Time `json:"reported_at"`
}
// Presence is one running service instance. Ready and Detail summarise its
// readiness probes; Detail names the degraded ones.
type Presence struct {
Checks []health.Component `json:"checks"`
Service string `json:"service"`
InstanceID string `json:"instance_id"`
Version string `json:"version"`
InstallType string `json:"install_type"`
StartedAt time.Time `json:"started_at"`
ReportedAt time.Time `json:"reported_at"`
Ready bool `json:"ready"`
Detail string `json:"detail,omitempty"`
}
// WorkflowStat is one row of the per-workflow metrics table.
type WorkflowStat struct {
Workflow string `json:"workflow"`
Runs int `json:"runs"`
Merged int `json:"merged"`
Completed int `json:"completed"`
PROpen int `json:"pr_open"`
Parked int `json:"parked"`
AvgTokens int `json:"avg_tokens"`
AvgSteps float64 `json:"avg_steps"`
TotalToken int `json:"total_tokens"`
// Running and LastStatus describe the workflow now: how many of its runs
// are working, and how its newest one ended.
Running int `json:"running"`
LastStatus string `json:"last_status"`
}
// StageStat is average stage duration and failure counts per stage.
type StageStat struct {
Workflow string `json:"workflow"`
Stage string `json:"stage"`
Runs int `json:"runs"`
AvgMs int `json:"avg_ms"`
Errors int `json:"errors"`
}
// DayTokens is token spend per UTC day.
type DayTokens struct {
Day string `json:"day"`
Tokens int `json:"tokens"`
}
// Sentinel errors, part of the wire contract: staterpc's
// mapError/unmapError match these by (code, canonical message) across the
// gRPC boundary, so their message strings are load-bearing and must never
// change.
var (
// ErrStaleTransition is returned when a task transition's expected
// from status does not match the task's current status.
ErrStaleTransition = errors.New("store: stale transition: task status does not match expected from status")
// ErrIllegalTransition is returned when the transition table does not allow
// the status pair.
ErrIllegalTransition = errors.New("store: illegal transition: status pair is not in the transition table")
// ErrInvalidStep is returned when a step-execution request names a kind the
// store does not record or a step whose identifying fields do not resolve.
// Distinct from the transition sentinels, which cover a row's state.
ErrInvalidStep = errors.New("store: invalid step execution request")
// ErrResumeIncomplete is returned when a retry resumes after a stage whose
// last run did not complete, so there are no results to start from.
ErrResumeIncomplete = errors.New("store: the step before the resume point did not complete in its last run")
// ErrBindingNotFound is returned when a binding ID does not exist.
ErrBindingNotFound = errors.New("store: binding not found")
// ErrBindingOverlap is returned when a binding's matcher overlaps an
// existing binding for the same source.
ErrBindingOverlap = errors.New("store: binding overlaps existing binding for source")
// ErrBindingTransition is returned when a binding state transition is
// rejected by the draft -> pending_approval -> armed machine.
ErrBindingTransition = errors.New("store: binding state transition rejected")
// ErrAlreadyDispatched is returned when a dispatch ledger already holds the
// key.
ErrAlreadyDispatched = errors.New("store: binding already dispatched for capture")
// ErrCallNotYours is returned when a workflow.call status read names a
// task that is not a callee of the calling task, or the caller itself
// does not exist: a caller reads only the tasks it started
ErrCallNotYours = errors.New("store: task is not a callee of the calling task")
// ErrCallCallerNotRunning is returned when a workflow.call enqueue names
// a caller that is not running: a stale retry, or a bug in the caller.
ErrCallCallerNotRunning = errors.New("store: workflow.call caller is not running")
// ErrCallDepthExceeded is returned when a workflow.call enqueue would
// pass workflow.MaxCallDepth. The engine refuses the same call first;
// the store re-checks because it owns the table.
ErrCallDepthExceeded = errors.New("store: workflow.call would pass the depth limit")
// ErrMappingNotFound is returned when a mapping ID does not exist.
ErrMappingNotFound = errors.New("store: mapping not found")
// ErrEventTypeNotFound is returned when an event type ID does not exist.
ErrEventTypeNotFound = errors.New("store: event type not found")
// ErrSourceNotFound is returned when a source path does not exist.
ErrSourceNotFound = errors.New("store: source not found")
// ErrSourceInUse is returned when deleting a source an armed binding
// fires on.
ErrSourceInUse = errors.New("store: source has an armed binding")
// ErrSourcePathTaken is returned when a new source's path is in use.
ErrSourcePathTaken = errors.New("store: source path already taken")
// ErrSourceSigningStale is returned when a signing write's expected
// from state does not match the stored one.
ErrSourceSigningStale = errors.New("store: source signing does not match expected state")
// ErrHarnessSecretNotFound is returned when no OAuth token set is
// stored for an org/service pair -- the setup terminal has not
// captured one yet.
ErrHarnessSecretNotFound = errors.New("store: harness secret not found")
// ErrRereviewCapReached is returned when a review gate re-review would
// round past prreview.MaxRereviewRounds. The task stays waiting and its
// gate row is unchanged. The dashboard answers it 409, like every other
// conflict.
ErrRereviewCapReached = errors.New("store: re-review cap reached")
// ErrResourceNotFound is returned when a control-plane resource kind has
// no stored document. It is in-process only (not on the gRPC wire): the
// control plane runs in the daemon against the local store.
ErrResourceNotFound = errors.New("resource not found")
// ErrResourceVersionConflict is returned when a resource write carries an
// expected version that does not match the stored revision.
ErrResourceVersionConflict = errors.New("resource version conflict")
)
// ErrUnavailable means the store could not serve a request now: it is
// restarting or unreachable. It is not paired with a canonical message the way
// the sentinels above are -- staterpc derives it from the Unavailable code
// alone, whose text varies by transport -- and it is the one store error a
// caller with budget may answer by retrying rather than by failing its own
// work. The wrapping keeps the gRPC status code readable through the chain.
var ErrUnavailable = errors.New("store: state store unavailable")
// Resource is one control-plane resource document: the stored value for an
// org's kind, its revision, and the audit record written with the last write.
// Resource and ResourceWrite live here so the UI-side consumers can reference
// them without linking the PostgreSQL implementation.
type Resource struct {
OrgID string
Kind string
Value []byte
Version int64
Actor string
Source string
RequestID string
ExpectedVersion int64
CurrentVersion int64
At time.Time
}
// DefaultOrgID is the org every resource written before orgs existed belongs
// to, and the org a request that names none acts in.
const DefaultOrgID = "org-sys"
// ResourceAuditKey is the sys_audit record key of one org's resource kind. The
// default org keeps the bare kind, so audit rows written before resources were
// per-org still resolve.
func ResourceAuditKey(orgID, kind string) string {
if orgID == DefaultOrgID {
return kind
}
return orgID + "/" + kind
}
// ResourceWrite is a control-plane resource write request. ExpectedVersion is
// the optimistic-concurrency guard: a write whose expectation does not match
// the stored revision is refused with ErrResourceVersionConflict.
type ResourceWrite struct {
OrgID string
Kind string
Value []byte
Actor string
Source string
RequestID string
ExpectedVersion int64
At time.Time
}
// ErrRunCredentialUnknown is a run credential that was never issued, was
// revoked, or has expired.
var ErrRunCredentialUnknown = errors.New("run credential unknown or expired")
// RunCredentialStore keeps run credentials by digest: the SHA-256 of the
// token a task's container presents. A lookup of an unknown or expired
// digest reports ErrRunCredentialUnknown.
type RunCredentialStore interface {
PutRunCredential(ctx context.Context, digest [32]byte, taskID int64, expires time.Time) error
DeleteRunCredential(ctx context.Context, digest [32]byte) error
RunCredentialTask(ctx context.Context, digest [32]byte, now time.Time) (int64, error)
DeleteExpiredRunCredentials(ctx context.Context, now time.Time) error
}
+198
View File
@@ -0,0 +1,198 @@
package storecontract
import (
"context"
"io"
"time"
"github.com/samcharles93/archie-core/internal/domain/workflow/task"
"github.com/samcharles93/archie-core/internal/events"
"github.com/samcharles93/archie-core/internal/logging"
)
// TaskLifecycle manages the core task state machine and enqueuing.
// Binding-triggered task creation lives on BindingTaskCreator, not here,
// so the lifecycle surface stays narrow and non-binding consumers (forge
// poll, chat spawn, drain loop) do not acquire a binding-specific shape.
type TaskLifecycle interface {
EnqueueIssue(ctx context.Context, owner, repo string, number int, title, body, labels, identity string) (bool, error)
EnqueueChatTask(ctx context.Context, owner, repo, title, body, wf, identity, origin, draft string, inputs map[string]any) (*task.Task, error)
ClaimNext(ctx context.Context) (*task.Task, error)
ClaimByIssue(ctx context.Context, owner, repo string, number int) (*task.Task, error)
Transition(ctx context.Context, taskID int64, from, to, detail string) error
Update(ctx context.Context, t *task.Task) error
Requeue(ctx context.Context, taskID int64, fromStatus, workflow string) error
RecoverStale(ctx context.Context) (int64, error)
}
// TaskParker parks a task with a park class and the blocker the park site
// knows, if any. Unknown classes become needs_decision.
type TaskParker interface {
ParkTask(ctx context.Context, taskID int64, from, detail, class, blocker string) error
}
// StepRecorder records a running execution's steps.
type StepRecorder interface {
StartStep(ctx context.Context, s task.StepStart) (int64, events.Event, error)
FinishStep(ctx context.Context, s task.StepFinish) (events.Event, error)
}
// ExecutionCanceller cancels an execution and its current attempt's open
// steps in one transaction. It also cancels the callees the execution was
// waiting on, and returns their ids (nested callees included) so the caller
// can stop their runs too.
type ExecutionCanceller interface {
CancelExecution(ctx context.Context, taskID int64, reason, to string) ([]int64, error)
}
// StepReader lists one execution's recorded steps.
type StepReader interface {
ListSteps(ctx context.Context, executionID int64, attempt int) ([]task.StepExecution, error)
}
// TaskArchiver removes one terminal task's local record with an optimistic
// status guard. It is separate from the already broad lifecycle contract so
// consumers that only run tasks do not acquire an operator-only capability.
type TaskArchiver interface {
OutcomePoster
ArchiveTask(ctx context.Context, taskID int64, fromStatus string, audit events.Event) (eventID int64, err error)
ClearTerminalTasks(ctx context.Context) (int64, error)
}
// TaskRetryer atomically requeues recoverable work and accounts for the new
// attempt so a partial write cannot evade the retry cap. retryMode is the
// operator's worktree choice for the next dispatch (taskstate.RetryMode),
// persisted on the row so the daemon reads it instead of inferring it. resume
// is where the attempt starts; a resume whose After stage did not complete in
// its last run is refused with ErrResumeIncomplete.
type TaskRetryer interface {
RetryTask(ctx context.Context, taskID int64, fromStatus, workflow, retryMode string, resume task.Resume, currentDefinition bool) error
}
// ReviewGateResponder records the operator's review-gate answer. A re-review
// increments rereview_rounds and requeues; one past the cap returns
// ErrRereviewCapReached and writes nothing.
type ReviewGateResponder interface {
RespondReviewGate(ctx context.Context, taskID int64, fromStatus, gate string, rereview bool, maxRounds int) error
}
// RemediationStarter queues a review-triggered remediation for an Archie-owned
// open-PR task: one guarded transition that carries the JSON-encoded review
// unit in the same write, so a claimed remediation always has its input. Runs
// for one task are serialised, so a distinct review that arrives while a
// remediation owns the task is queued behind it and remediated once that run
// finishes; a re-delivered unit is absorbed, not queued twice.
// UpdateReviewPayload merges late-arriving comments into the unclaimed unit
// for the same review, active or queued behind a running remediation.
type RemediationStarter interface {
BeginRemediation(ctx context.Context, taskID int64, payload string) error
UpdateReviewPayload(ctx context.Context, taskID int64, payload string) error
// SetReviewCursors persists the poll backstop's per-task review and
// comment high-water marks (pr-review-remediation.md decision 2). The
// pr_open guard is part of the contract: cursors only move while no
// remediation run owns the task.
SetReviewCursors(ctx context.Context, taskID, reviewCursor, commentCursor int64) error
}
// OutcomePoster tracks which finished chat tasks have reported back to their
// conversation.
type OutcomePoster interface {
// UnpostedOutcomes returns finished chat tasks whose outcome has not been
// posted to the conversation that created them.
UnpostedOutcomes(ctx context.Context) ([]*task.Task, error)
// MarkOutcomePosted records that the task's state at status was posted.
MarkOutcomePosted(ctx context.Context, taskID int64, status string) error
}
// TaskQueries groups read-only task accessors.
type TaskQueries interface {
TaskByIssue(ctx context.Context, owner, repo string, number int) (*task.Task, error)
OpenTaskByPR(ctx context.Context, owner, repo string, number int) (*task.Task, error)
TaskByID(ctx context.Context, taskID int64) (*task.Task, error)
OpenPRs(ctx context.Context) ([]task.Task, error)
// ActiveTasksByOrigin returns the queued and running tasks the
// conversation origin created, for its /stop.
ActiveTasksByOrigin(ctx context.Context, origin string) ([]*task.Task, error)
// Tasks returns the most recently updated tasks.
Tasks(ctx context.Context, limit int) ([]task.Task, error)
// TasksPage returns one page of tasks, most recently updated first.
TasksPage(ctx context.Context, page TaskPage) ([]task.Task, error)
StatusCounts(ctx context.Context) (map[string]int, error)
}
// TaskEvents groups observability and lifecycle methods.
type TaskEvents interface {
InsertEvent(ctx context.Context, e events.Event) (int64, error)
EventsSince(ctx context.Context, cursor string, limit int) ([]events.Event, error)
// TaskEvents returns a task's timeline, without agent transcripts.
TaskEvents(ctx context.Context, taskID int64) ([]events.Event, error)
// TaskTranscript returns one attempt's agent transcript entries in order.
TaskTranscript(ctx context.Context, taskID int64, attempt int) ([]events.Event, error)
WorkflowStats(ctx context.Context) ([]WorkflowStat, error)
StageStats(ctx context.Context) ([]StageStat, error)
TokensByDay(ctx context.Context, days int) ([]DayTokens, error)
Close() error
}
// TaskPage selects one page of the task list. Empty Statuses matches every
// status. A zero After starts at the newest task; otherwise the page begins
// after the task it names, in (updated_at, id) descending order.
type TaskPage struct {
Statuses []string
// Workflow, when set, keeps only that workflow's tasks.
Workflow string
After TaskCursor
Limit int
}
// TaskCursor is the position of the last task a page returned.
type TaskCursor struct {
UpdatedAt time.Time
ID int64
}
// WorkflowCaller starts a workflow.call callee and reads it back.
type WorkflowCaller = task.Caller
// TaskLogStore reads one task attempt's log. Found=false means no log;
// ErrTaskLogsUnavailable means this process cannot read logs.
type TaskLogStore interface {
// TaskLog returns a page of one attempt's decoded log. attempt 0 selects
// the task's current attempt.
TaskLog(ctx context.Context, taskID int64, attempt int, q logging.Query) (logging.TaskLogPage, error)
// TaskLogContent writes one attempt's log verbatim to w, for a download.
// found is false, with a nil error, when the attempt has no log file.
TaskLogContent(ctx context.Context, taskID int64, attempt int, w io.Writer) (found bool, err error)
}
// WorkflowStat is one row of the per-workflow metrics table.
type WorkflowStat struct {
Workflow string `json:"workflow"`
Runs int `json:"runs"`
Merged int `json:"merged"`
Completed int `json:"completed"`
PROpen int `json:"pr_open"`
Parked int `json:"parked"`
AvgTokens int `json:"avg_tokens"`
AvgSteps float64 `json:"avg_steps"`
TotalToken int `json:"total_tokens"`
// Running and LastStatus describe the workflow now: how many of its runs
// are working, and how its newest one ended.
Running int `json:"running"`
LastStatus string `json:"last_status"`
}
// StageStat is average stage duration and failure counts per stage.
type StageStat struct {
Workflow string `json:"workflow"`
Stage string `json:"stage"`
Runs int `json:"runs"`
AvgMs int `json:"avg_ms"`
Errors int `json:"errors"`
}
// DayTokens is token spend per UTC day.
type DayTokens struct {
Day string `json:"day"`
Tokens int `json:"tokens"`
}
+3 -4
View File
@@ -36,8 +36,8 @@ type handoffSettings struct {
Detail string `yaml:"detail" doc:"Why the task was handed off."`
}
// newHandoffStage requeues the task under another workflow: the same task,
// not a child run, so its history stays in one place.
// newHandoffStage completes the run and starts a linked run of another
// workflow on the same work item.
func newHandoffStage(settings yaml.Node) (Stage, error) {
s, err := decodeSettings[handoffSettings](HandoffStepName, settings)
if err != nil {
@@ -47,8 +47,7 @@ func newHandoffStage(settings yaml.Node) (Stage, error) {
return Stage{}, fmt.Errorf("%s: settings.workflow is required", HandoffStepName)
}
return Stage{Name: HandoffStepName, Run: func(_ context.Context, tc *TaskContext) error {
tc.Task.Workflow = strings.TrimSpace(s.Workflow)
tc.Outcome = Outcome{Status: StatusQueued, Detail: s.Detail}
tc.Outcome = Outcome{Status: StatusCompleted, Detail: s.Detail, Handoff: strings.TrimSpace(s.Workflow)}
return nil
}}, nil
}
+1 -1
View File
@@ -113,7 +113,7 @@ func runRepoTooling(ctx context.Context, tc *TaskContext) error {
}
type repoCommitSettings struct {
Message string `yaml:"message" doc:"The commit message."`
Message string `yaml:"message" doc:"The commit message. Hooks in the checkout never run, so commit-message policy (commitlint, DCO) is not applied."`
// Reference is the verb that links the commit to the task's issue
// ("Fixes", "Implements", "Refs"); empty links nothing.
Reference string `yaml:"reference" enum:"Fixes,Implements,Refs" doc:"How the commit links to the task issue."`
+4
View File
@@ -16,4 +16,8 @@ type Caller interface {
// call_parent_task_id is callerTaskID, so a caller reads only its own
// callees.
CallStatus(ctx context.Context, callerTaskID, callTaskID int64) (status, detail string, outputs map[string]any, err error)
// Handoff completes the running task fromTaskID with detail and starts
// its linked successor on the same work item under workflow, returning
// the successor.
Handoff(ctx context.Context, fromTaskID int64, workflow, detail string) (*Task, error)
}
+4
View File
@@ -141,6 +141,10 @@ type Task struct {
// one; 0 on a root run. A caller reads only the tasks it started,
// through WorkflowCallStatus.
CallParentTaskID int64 `json:"call_parent_task_id,omitempty"`
// HandoffFromTaskID is the run whose workflow.handoff started this one,
// and HandoffToTaskID the run this one handed off to; 0 when none.
HandoffFromTaskID int64 `json:"handoff_from_task_id,omitempty"`
HandoffToTaskID int64 `json:"handoff_to_task_id,omitempty"`
// CallDepth counts workflow.call hops from the root run. The engine
// refuses a call past workflow.MaxCallDepth, and the store re-checks on
// insert, so no enqueue path can pass the limit.
+5
View File
@@ -45,6 +45,11 @@ type recordedForge struct {
deliveries Deliveries
}
func (f recordedForge) CreateIssue(ctx context.Context, owner, repo, title, _ string, _ []string) (int, error) {
f.deliveries.Record(ctx, fmt.Sprintf("open issue %q in %s/%s", title, owner, repo))
return 0, nil
}
func (f recordedForge) CloseIssue(ctx context.Context, owner, repo string, number int, _ string) error {
f.deliveries.Record(ctx, fmt.Sprintf("close issue %s/%s#%d", owner, repo, number))
return nil
+27
View File
@@ -279,6 +279,9 @@ func transcriptData(entry agentrun.TranscriptEntry) (map[string]any, error) {
type Outcome struct {
Status string // Status* value; empty = continue to next stage
Detail string // park reason / close rationale / PR body context
// Handoff names the workflow a linked successor run starts under once
// this run completes; empty starts none.
Handoff string
}
// Stage is one named step. Returning an error parks the task with the
@@ -507,6 +510,26 @@ func publishEvent(ctx context.Context, tc *TaskContext, e events.Event) {
tc.Bus.Publish(ctx, e)
}
// handoff completes the run and starts its linked successor in one store
// write, so the work item keeps exactly one current run.
func handoff(ctx context.Context, tc *TaskContext, log *slog.Logger) {
if tc.Calls == nil {
park(ctx, tc, "workflow.handoff: this runner cannot start a linked run")
return
}
next, err := tc.Calls.Handoff(ctx, tc.Task.ID, tc.Outcome.Handoff, tc.Outcome.Detail)
if err != nil {
log.ErrorContext(ctx, "workflow handoff failed", "err", err)
park(ctx, tc, fmt.Sprintf("workflow.handoff to %s: %v", tc.Outcome.Handoff, err))
return
}
tc.Task.HandoffToTaskID = next.ID
tc.Emit(ctx, events.KindOutcome, tc.Stage, tc.Outcome.Detail, map[string]any{
"status": tc.Outcome.Status, "handoff_task_id": next.ID, "workflow": tc.Outcome.Handoff,
})
log.InfoContext(ctx, "workflow handed off", "workflow", tc.Outcome.Handoff, "task", next.ID)
}
func finish(ctx context.Context, tc *TaskContext, log *slog.Logger) {
t := tc.Task
if tc.Outcome.Status == StatusParked {
@@ -525,6 +548,10 @@ func finish(ctx context.Context, tc *TaskContext, log *slog.Logger) {
park(ctx, tc, fmt.Sprintf("persist workflow outcome: %v", err))
return
}
if tc.Outcome.Handoff != "" {
handoff(ctx, tc, log)
return
}
if err := tc.Store.Transition(ctx, t.ID, StatusRunning, tc.Outcome.Status, tc.Outcome.Detail); err != nil {
log.ErrorContext(ctx, "workflow outcome transition failed", "err", err)
return
@@ -37,6 +37,10 @@ func (c *callSpy) StartCall(_ context.Context, callerTaskID int64, callKey, _ st
return child, nil
}
func (c *callSpy) Handoff(context.Context, int64, string, string) (*task.Task, error) {
return nil, errors.New("no handoff")
}
func (c *callSpy) CallStatus(context.Context, int64, int64) (string, string, map[string]any, error) {
return StatusCompleted, "", nil, nil
}
@@ -384,6 +388,10 @@ func (c *wobblyCaller) StartCall(_ context.Context, callerTaskID int64, _, _ str
return &task.Task{ID: 42, CallParentTaskID: callerTaskID}, nil
}
func (c *wobblyCaller) Handoff(context.Context, int64, string, string) (*task.Task, error) {
return nil, errors.New("no handoff")
}
func (c *wobblyCaller) CallStatus(context.Context, int64, int64) (string, string, map[string]any, error) {
c.mu.Lock()
defer c.mu.Unlock()
+4
View File
@@ -38,6 +38,10 @@ type IssueForge interface {
// Comment posts an issue (or PR) comment and returns its id.
Comment(ctx context.Context, owner, repo string, number int, body string) (int64, error)
// CreateIssue opens an issue with title, body and optional labels,
// returning its number.
CreateIssue(ctx context.Context, owner, repo, title, body string, labels []string) (int, error)
// CloseIssue closes an issue with an optional final comment.
CloseIssue(ctx context.Context, owner, repo string, number int, comment string) error
+4
View File
@@ -30,6 +30,10 @@ func (n *NoopForge) Comment(ctx context.Context, owner, repo string, number int,
return 0, nil
}
func (n *NoopForge) CreateIssue(ctx context.Context, owner, repo, title, body string, labels []string) (int, error) {
return 0, nil
}
func (n *NoopForge) CloseIssue(ctx context.Context, owner, repo string, number int, comment string) error {
return nil
}
+43 -2
View File
@@ -23,6 +23,7 @@ import (
const (
SubjectComment = "archie.forge.comment"
SubjectCreateIssue = "archie.forge.create_issue"
SubjectCloseIssue = "archie.forge.close_issue"
SubjectCreatePR = "archie.forge.create_pr"
SubjectLinkBranch = "archie.forge.link_branch"
@@ -35,8 +36,9 @@ const (
// set a task of that identity may publish to.
func Subjects(identity string) []string {
bases := []string{
SubjectComment, SubjectCloseIssue, SubjectCreatePR, SubjectLinkBranch,
SubjectSetStateLabel, SubjectCreateReviewComment, SubjectReplyToReview,
SubjectComment, SubjectCreateIssue, SubjectCloseIssue, SubjectCreatePR,
SubjectLinkBranch, SubjectSetStateLabel, SubjectCreateReviewComment,
SubjectReplyToReview,
}
out := make([]string, len(bases))
for i, base := range bases {
@@ -80,6 +82,17 @@ type CommentResponse struct {
natsrpc.Envelope
}
type CreateIssueRequest struct {
Target
Title, Body string
Labels []string
}
type CreateIssueResponse struct {
Number int `json:"number"`
natsrpc.Envelope
}
type CloseIssueRequest struct {
Target
Number int
@@ -168,6 +181,7 @@ func (s *Server) RegisterFor(nc *nats.Conn, identity string) (unsubscribe func()
}
return natsrpc.RegisterAll(nc, []natsrpc.Registration{
{Subject: serverSubject(identity, SubjectComment), Handler: as(s.handleComment)},
{Subject: serverSubject(identity, SubjectCreateIssue), Handler: as(s.handleCreateIssue)},
{Subject: serverSubject(identity, SubjectCloseIssue), Handler: as(s.handleCloseIssue)},
{Subject: serverSubject(identity, SubjectCreatePR), Handler: as(s.handleCreatePR)},
{Subject: serverSubject(identity, SubjectLinkBranch), Handler: as(s.handleLinkBranch)},
@@ -249,6 +263,21 @@ func (s *Server) handleComment(ctx context.Context, identity string, msg *nats.M
s.respond(msg, CommentResponse{ID: id, Envelope: natsrpc.NewEnvelope(err)})
}
func (s *Server) handleCreateIssue(ctx context.Context, identity string, msg *nats.Msg) {
var req CreateIssueRequest
if err := json.Unmarshal(msg.Data, &req); err != nil {
s.respond(msg, CreateIssueResponse{Envelope: natsrpc.NewEnvelope(fmt.Errorf("decode create_issue request: %w", err))})
return
}
client, err := s.authorizedForge(ctx, identity, req.Target, "comment")
if err != nil {
s.respond(msg, CreateIssueResponse{Envelope: natsrpc.NewEnvelope(err)})
return
}
number, err := client.CreateIssue(ctx, req.Owner, req.Repo, req.Title, req.Body, req.Labels)
s.respond(msg, CreateIssueResponse{Number: number, Envelope: natsrpc.NewEnvelope(err)})
}
func (s *Server) handleCloseIssue(ctx context.Context, identity string, msg *nats.Msg) {
var req CloseIssueRequest
if err := json.Unmarshal(msg.Data, &req); err != nil {
@@ -392,6 +421,18 @@ func (c *Client) Comment(ctx context.Context, owner, repo string, number int, bo
return resp.ID, nil
}
func (c *Client) CreateIssue(ctx context.Context, owner, repo, title, body string, labels []string) (int, error) {
req := CreateIssueRequest{Target: c.target(owner, repo), Title: title, Body: body, Labels: labels}
resp, err := natsrpc.Call[CreateIssueResponse](ctx, c.rpc(), c.subject(SubjectCreateIssue), req)
if err != nil {
return 0, err
}
if err := resp.Err(); err != nil {
return 0, err
}
return resp.Number, nil
}
func (c *Client) CloseIssue(ctx context.Context, owner, repo string, number int, comment string) error {
req := CloseIssueRequest{Target: c.target(owner, repo), Number: number, Comment: comment}
resp, err := natsrpc.Call[Response](ctx, c.rpc(), c.subject(SubjectCloseIssue), req)
+129
View File
@@ -0,0 +1,129 @@
package gateway
import (
"context"
"fmt"
"strings"
"github.com/samcharles93/archie-core/internal/domain/access"
"github.com/samcharles93/archie-core/internal/forge"
"github.com/samcharles93/archie-core/internal/tools"
)
// maxIssueListed bounds what issue_list returns, so a large backlog does not
// land in the context window whole.
const maxIssueListed = 200
// IssueSource is the forge surface the issue tools use, reached through the
// daemon.
type IssueSource interface {
OpenIssues(ctx context.Context, owner, repo string) ([]forge.Issue, error)
CreateIssue(ctx context.Context, owner, repo, title, body string, labels []string) (int, error)
}
// IssueSummary is one open issue as issue_list shows it.
type IssueSummary struct {
Number int `json:"number"`
Title string `json:"title"`
Labels []string `json:"labels,omitempty"`
}
// IssueTools returns issue_list and issue_create for curators. Filing work
// feeds the daemon's issue intake, so both are checked as acting on tasks.
// Nil source or authorizer registers none.
func IssueTools(source IssueSource, authz access.Authorizer) []tools.ToolEntry {
if source == nil || authz == nil {
return nil
}
return []tools.ToolEntry{
guarded(authz, access.ActionRead, access.KindTask, nil, issueListTool(source)),
guarded(authz, access.ActionCreate, access.KindTask, nil, issueCreateTool(source)),
}
}
var repoProperty = map[string]any{
"type": "string",
"description": "The repository as owner/name, as run_list reports it.",
}
func splitRepo(input map[string]any) (owner, repo string, err error) {
raw, _ := input["repo"].(string)
owner, repo, ok := strings.Cut(strings.TrimSpace(raw), "/")
if !ok || owner == "" || repo == "" || strings.Contains(repo, "/") {
return "", "", fmt.Errorf("repo must be owner/name")
}
return owner, repo, nil
}
func issueListTool(source IssueSource) tools.ToolEntry {
return tools.ToolEntry{
Name: "issue_list",
Toolset: "issues",
Description: "List a repository's open issues (number, title, labels). Check it before filing so you never duplicate one.",
Classification: tools.ClassIdempotent,
Schema: tools.JSONSchema{
"type": "object",
"properties": map[string]any{"repo": repoProperty},
"required": []string{"repo"},
},
Handler: func(ctx context.Context, input map[string]any) (any, error) {
owner, repo, err := splitRepo(input)
if err != nil {
return nil, fmt.Errorf("issue_list: %w", err)
}
issues, err := source.OpenIssues(ctx, owner, repo)
if err != nil {
return nil, fmt.Errorf("issue_list: %w", err)
}
out := make([]IssueSummary, 0, min(len(issues), maxIssueListed))
for _, is := range issues[:min(len(issues), maxIssueListed)] {
out = append(out, IssueSummary{Number: is.Number, Title: is.Title, Labels: is.Labels})
}
return out, nil
},
}
}
func issueCreateTool(source IssueSource) tools.ToolEntry {
return tools.ToolEntry{
Name: "issue_create",
Toolset: "issues",
Description: "File a new issue in a repository. List open issues first and file only work no open issue covers. " +
"The body must cite the run that produced it.",
Classification: tools.ClassMutating,
Schema: tools.JSONSchema{
"type": "object",
"properties": map[string]any{
"repo": repoProperty,
"title": map[string]any{"type": "string"},
"body": map[string]any{"type": "string"},
"labels": map[string]any{"type": "array", "items": map[string]any{"type": "string"}},
},
"required": []string{"repo", "title", "body"},
},
Handler: func(ctx context.Context, input map[string]any) (any, error) {
owner, repo, err := splitRepo(input)
if err != nil {
return nil, fmt.Errorf("issue_create: %w", err)
}
title, _ := input["title"].(string)
body, _ := input["body"].(string)
if strings.TrimSpace(title) == "" || strings.TrimSpace(body) == "" {
return nil, fmt.Errorf("issue_create: title and body are required")
}
var labels []string
if raw, ok := input["labels"].([]any); ok {
for _, l := range raw {
if s, ok := l.(string); ok && s != "" {
labels = append(labels, s)
}
}
}
number, err := source.CreateIssue(ctx, owner, repo, title, body, labels)
if err != nil {
return nil, fmt.Errorf("issue_create: %w", err)
}
return map[string]any{"number": number}, nil
},
}
}
+137
View File
@@ -0,0 +1,137 @@
package gateway
import (
"context"
"fmt"
"time"
"github.com/samcharles93/archie-core/internal/domain/access"
"github.com/samcharles93/archie-core/internal/domain/workflow/task"
"github.com/samcharles93/archie-core/internal/events"
"github.com/samcharles93/archie-core/internal/tools"
)
// RunSource is the state-store surface the run-inspection tools use. The
// State Store client satisfies it directly.
type RunSource interface {
// Tasks returns the most recently updated tasks.
Tasks(ctx context.Context, limit int) ([]task.Task, error)
// TaskEvents returns one task's events, oldest first.
TaskEvents(ctx context.Context, taskID int64) ([]events.Event, error)
}
// RunSummary is one task as run_list shows it.
type RunSummary struct {
ID int64 `json:"id"`
Repo string `json:"repo"`
Title string `json:"title"`
Status string `json:"status"`
Workflow string `json:"workflow,omitempty"`
Issue int `json:"issue,omitempty"`
PRNumber int `json:"pr_number,omitempty"`
ParkReason string `json:"park_reason,omitempty"`
UpdatedAt time.Time `json:"updated_at"`
}
// RunEvent is one task event as run_events shows it.
type RunEvent struct {
At time.Time `json:"at"`
Kind string `json:"kind"`
Stage string `json:"stage,omitempty"`
Detail string `json:"detail,omitempty"`
}
// RunTools returns the read-only run-inspection tools for long-running
// curators: what ran recently, and each run's event trail. Reads are checked
// as reading tasks, the same check the dashboard's task list gets. Nil
// source or authorizer registers none.
func RunTools(source RunSource, authz access.Authorizer) []tools.ToolEntry {
if source == nil || authz == nil {
return nil
}
read := func(entry tools.ToolEntry) tools.ToolEntry {
return guarded(authz, access.ActionRead, access.KindTask, nil, entry)
}
return []tools.ToolEntry{
read(runListTool(source)),
read(runEventsTool(source)),
}
}
func runListTool(source RunSource) tools.ToolEntry {
return tools.ToolEntry{
Name: "run_list",
Toolset: "runs",
Description: "List the most recently updated task runs across the instance, most recent first. " +
"Use it to survey what ran, what parked and why.",
Classification: tools.ClassIdempotent,
Schema: tools.JSONSchema{
"type": "object",
"properties": map[string]any{
"limit": map[string]any{
"type": "integer",
"description": fmt.Sprintf(
"Maximum runs to return. Defaults to %d, capped at %d.",
defaultTaskListLimit, maxTaskListLimit,
),
},
},
},
Handler: func(ctx context.Context, input map[string]any) (any, error) {
rows, err := source.Tasks(ctx, taskListLimit(input))
if err != nil {
return nil, fmt.Errorf("run_list: %w", err)
}
out := make([]RunSummary, 0, len(rows))
for _, t := range rows {
out = append(out, RunSummary{
ID: t.ID,
Repo: t.Owner + "/" + t.Repo,
Title: t.Title,
Status: t.Status,
Workflow: t.Workflow,
Issue: t.IssueNumber,
PRNumber: t.PRNumber,
ParkReason: t.ParkReason,
UpdatedAt: t.UpdatedAt,
})
}
return out, nil
},
}
}
func runEventsTool(source RunSource) tools.ToolEntry {
return tools.ToolEntry{
Name: "run_events",
Toolset: "runs",
Description: "List one task run's event trail, oldest first: stage transitions, deliveries, " +
"failures and park reasons. Use it to see why a run ended the way it did.",
Classification: tools.ClassIdempotent,
Schema: tools.JSONSchema{
"type": "object",
"properties": map[string]any{
"task_id": map[string]any{
"type": "integer",
"description": "The task run's ID, as run_list reports it.",
},
},
"required": []string{"task_id"},
},
Handler: func(ctx context.Context, input map[string]any) (any, error) {
id, ok := asInt64(input["task_id"])
if !ok || id <= 0 {
return nil, fmt.Errorf("run_events: task_id must be a positive integer")
}
evs, err := source.TaskEvents(ctx, id)
if err != nil {
return nil, fmt.Errorf("run_events: %w", err)
}
out := make([]RunEvent, 0, len(evs))
for _, e := range evs {
out = append(out, RunEvent{At: e.At, Kind: e.Kind, Stage: e.Stage, Detail: e.Detail})
}
return out, nil
},
}
}
@@ -112,3 +112,84 @@ func TestWorkspaceEnvironment(t *testing.T) {
})
}
}
// TestCanaryGrant pins the scoped canary grant: the one identity it names may
// read the signal surface, read and write update-settings, and trigger an
// install -- and nothing else, neither for it nor for anyone else. The canary
// principal carries the developer role (API-created service accounts are
// assigned it); the grant's forbid must keep the role from widening access.
func TestCanaryGrant(t *testing.T) {
const canaryID = identity.IdentityID("canary-svc")
grant := access.CanaryGrantPolicy(org.DefaultOrgID, string(canaryID))
forbid := access.CanaryGrantForbidPolicy(org.DefaultOrgID, string(canaryID))
engine := New(append(access.ShippedOrgPolicies(org.DefaultOrgID), grant, forbid))
if len(engine.Problems()) != 0 {
t.Fatalf("problems = %v, want the canary grant to compile", engine.Problems())
}
canary := access.Principal{
IdentityID: canaryID, Org: org.DefaultOrgID,
Memberships: []org.Membership{{IdentityID: canaryID, OrgID: org.DefaultOrgID, Role: org.RoleDeveloper}},
}
other := access.Principal{IdentityID: "someone-else", Org: org.DefaultOrgID}
at := func(kind access.ResourceKind, id string) access.Resource {
return access.Resource{Kind: kind, ID: id, Org: org.DefaultOrgID}
}
tests := []struct {
name string
principal access.Principal
action access.Action
resource access.Resource
allowed bool
}{
{"read the dashboard", canary, access.ActionRead, at(access.KindDashboard, ""), true},
{"read tasks", canary, access.ActionRead, at(access.KindTask, ""), true},
{"read one task", canary, access.ActionRead, at(access.KindTask, "7"), true},
{"read update-settings", canary, access.ActionRead, at(access.KindPolicy, "update-settings"), true},
{"install an update", canary, access.ActionUpdate, at(access.KindDashboard, "update"), true},
{"replace update-settings", canary, access.ActionManageIdentities, at(access.KindPolicy, "update-settings"), true},
{"read a secret", canary, access.ActionReadSecret, at(access.KindSecret, "github-token"), false},
{"delete a task", canary, access.ActionDelete, at(access.KindTask, "7"), false},
{"manage policies", canary, access.ActionManagePolicies, at(access.KindPolicy, "org-role-read"), false},
{"run a workflow", canary, access.ActionRun, at(access.KindWorkflow, "triage"), false},
{"create anything", canary, access.ActionCreate, at(access.KindTask, ""), false},
{"update another object", canary, access.ActionUpdate, at(access.KindDashboard, "other"), false},
{"read another object", canary, access.ActionRead, at(access.KindPolicy, "other-policy"), false},
{"manage identities on another object", canary, access.ActionManageIdentities, at(access.KindIdentity, "someone-else"), false},
{"read logs", canary, access.ActionReadLogs, at(access.KindTask, "7"), false},
{"read the audit feed", canary, access.ActionReadAudit, at(access.KindAudit, ""), false},
{"a resource of another org", canary, access.ActionRead, access.Resource{Kind: access.KindDashboard, Org: "acme"}, false},
{"another identity reads the dashboard", other, access.ActionRead, at(access.KindDashboard, ""), false},
{"another identity installs an update", other, access.ActionUpdate, at(access.KindDashboard, "update"), false},
{"another identity replaces update-settings", other, access.ActionManageIdentities, at(access.KindPolicy, "update-settings"), false},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
got := engine.Authorize(tt.principal, tt.action, tt.resource, access.Context{})
if got.Allowed != tt.allowed {
t.Fatalf("allowed = %v, want %v (%+v)", got.Allowed, tt.allowed, got)
}
})
}
}
// TestCanaryGrantValidates pins that the canary grant passes the domain's
// structural check and the engine's schema validation, so it can be installed
// through the policy store.
func TestCanaryGrantValidates(t *testing.T) {
grant := access.CanaryGrantPolicy(org.DefaultOrgID, "canary-svc")
forbid := access.CanaryGrantForbidPolicy(org.DefaultOrgID, "canary-svc")
for _, p := range []access.Policy{grant, forbid} {
if err := p.Validate(); err != nil {
t.Fatalf("Validate %s: %v", p.ID, err)
}
}
engine := New([]access.Policy{grant, forbid})
for _, p := range []access.Policy{grant, forbid} {
if err := engine.Validate(p); err != nil {
t.Fatalf("engine Validate %s: %v", p.ID, err)
}
}
if len(engine.Problems()) != 0 {
t.Fatalf("problems = %v, want none", engine.Problems())
}
}
@@ -193,6 +193,12 @@ func (d deadlineCaller) StartCall(ctx context.Context, callerTaskID int64, callK
return d.Caller.StartCall(ctx, callerTaskID, callKey, wf, inputs)
}
func (d deadlineCaller) Handoff(ctx context.Context, fromTaskID int64, wf, detail string) (*task.Task, error) {
ctx, cancel := d.withDeadline(ctx)
defer cancel()
return d.Caller.Handoff(ctx, fromTaskID, wf, detail)
}
func (d deadlineCaller) CallStatus(ctx context.Context, callerTaskID, callTaskID int64) (string, string, map[string]any, error) {
ctx, cancel := d.withDeadline(ctx)
defer cancel()
@@ -52,3 +52,13 @@ func DefaultUpdateStateDir() string {
}
return filepath.Join(dir, InstanceName(), "update")
}
// UpdateStateDir is the one resolution of the host's update state directory:
// ARCHIE_UPDATE_STATE_DIR when set, else the default. The daemon's
// transactions and the dashboard's update snapshot must read the same one.
func UpdateStateDir() string {
if dir := os.Getenv("ARCHIE_UPDATE_STATE_DIR"); dir != "" {
return dir
}
return DefaultUpdateStateDir()
}
@@ -195,6 +195,16 @@ func (f *Forge) Comment(ctx context.Context, owner, repo string, number int, bod
return resp.GetId(), err
}
func (f *Forge) CreateIssue(ctx context.Context, owner, repo, title, body string, labels []string) (int, error) {
resp, err := call(ctx, f, "create issue", func(c forgev1.ForgeServiceClient) (*forgev1.CreateIssueResponse, error) {
return c.CreateIssue(ctx, &forgev1.CreateIssueRequest{Repo: ref(owner, repo), Title: title, Body: body, Labels: labels})
})
if err != nil {
return 0, err
}
return int(resp.GetNumber()), nil
}
func (f *Forge) CloseIssue(ctx context.Context, owner, repo string, number int, comment string) error {
_, err := call(ctx, f, "close issue", func(c forgev1.ForgeServiceClient) (*forgev1.CloseIssueResponse, error) {
return c.CloseIssue(ctx, &forgev1.CloseIssueRequest{Repo: ref(owner, repo), Number: int32(number), Comment: comment}) // issue numbers fit int32
@@ -0,0 +1,123 @@
// Package issuerpc lets the Gateway's curators list and file issues through
// the daemon, which alone holds forge credentials.
package issuerpc
import (
"context"
"encoding/json"
"errors"
"fmt"
"log/slog"
"time"
"github.com/nats-io/nats.go"
"github.com/samcharles93/archie-core/internal/forge"
"github.com/samcharles93/archie-core/internal/natsrpc"
)
const (
listSubject = "archie.gateway.issue-list"
createSubject = "archie.gateway.issue-create"
)
type request struct {
Owner string `json:"owner"`
Repo string `json:"repo"`
Title string `json:"title,omitempty"`
Body string `json:"body,omitempty"`
Labels []string `json:"labels,omitempty"`
}
type listResponse struct {
natsrpc.Envelope
Issues []forge.Issue `json:"issues,omitempty"`
}
type createResponse struct {
natsrpc.Envelope
Number int `json:"number,omitempty"`
}
// Server is the daemon side. Forge returns the instance forge, or nil while
// none is configured; Allowed refuses a repository the instance does not manage.
type Server struct {
Forge func() forge.IssueForge
Allowed func(ctx context.Context, owner, repo string) error
Log *slog.Logger
}
func (s *Server) forge(ctx context.Context, req request) (forge.IssueForge, error) {
if err := s.Allowed(ctx, req.Owner, req.Repo); err != nil {
return nil, err
}
f := s.Forge()
if f == nil {
return nil, errors.New("forge is not configured")
}
return f, nil
}
// Register serves the issue subjects on nc.
func (s *Server) Register(nc *nats.Conn) (func(), error) {
return natsrpc.RegisterAll(nc, []natsrpc.Registration{
{Subject: listSubject, Handler: func(ctx context.Context, msg *nats.Msg) {
var req request
if err := json.Unmarshal(msg.Data, &req); err != nil {
natsrpc.Respond(msg, s.Log, "issuerpc", listResponse{Envelope: natsrpc.NewEnvelope(err)})
return
}
f, err := s.forge(ctx, req)
var issues []forge.Issue
if err == nil {
issues, err = f.IssuesWithLabel(ctx, req.Owner, req.Repo, "")
}
natsrpc.Respond(msg, s.Log, "issuerpc", listResponse{Envelope: natsrpc.NewEnvelope(err), Issues: issues})
}},
{Subject: createSubject, Handler: func(ctx context.Context, msg *nats.Msg) {
var req request
if err := json.Unmarshal(msg.Data, &req); err != nil {
natsrpc.Respond(msg, s.Log, "issuerpc", createResponse{Envelope: natsrpc.NewEnvelope(err)})
return
}
f, err := s.forge(ctx, req)
var number int
if err == nil {
number, err = f.CreateIssue(ctx, req.Owner, req.Repo, req.Title, req.Body, req.Labels)
}
natsrpc.Respond(msg, s.Log, "issuerpc", createResponse{Envelope: natsrpc.NewEnvelope(err), Number: number})
}},
})
}
// Client is the Gateway side.
type Client struct {
Conn *nats.Conn
Timeout time.Duration
}
func (c Client) rpc() *natsrpc.Client { return &natsrpc.Client{Conn: c.Conn, Timeout: c.Timeout} }
// OpenIssues lists every open issue (not PR) of owner/repo.
func (c Client) OpenIssues(ctx context.Context, owner, repo string) ([]forge.Issue, error) {
resp, err := natsrpc.Call[listResponse](ctx, c.rpc(), listSubject, request{Owner: owner, Repo: repo})
if err != nil {
return nil, err
}
if err := resp.Err(); err != nil {
return nil, fmt.Errorf("list issues: %w", err)
}
return resp.Issues, nil
}
// CreateIssue files an issue and returns its number.
func (c Client) CreateIssue(ctx context.Context, owner, repo, title, body string, labels []string) (int, error) {
resp, err := natsrpc.Call[createResponse](ctx, c.rpc(), createSubject, request{Owner: owner, Repo: repo, Title: title, Body: body, Labels: labels})
if err != nil {
return 0, err
}
if err := resp.Err(); err != nil {
return 0, fmt.Errorf("create issue: %w", err)
}
return resp.Number, nil
}
+7 -2
View File
@@ -31,11 +31,16 @@ import (
// the tool_calls transcript the composition projection writes through.
var (
_ storecontract.CaptureStore = (*EDA)(nil)
_ storecontract.CaptureRefusalStore = (*EDA)(nil)
_ storecontract.MappingStore = (*EDA)(nil)
_ storecontract.BindingStore = (*EDA)(nil)
_ storecontract.BindingDispatcher = (*EDA)(nil)
_ storecontract.MappingMatchRecorder = (*EDA)(nil)
_ storecontract.EventTypeStore = (*EDA)(nil)
_ storecontract.BindingStore = (*EDA)(nil)
_ storecontract.SourceStore = (*EDA)(nil)
_ storecontract.DispatchLedger = (*EDA)(nil)
_ storecontract.BindingDispatcher = (*EDA)(nil)
_ storecontract.PlaybookDispatcher = (*EDA)(nil)
_ storecontract.HarnessSecretStore = (*EDA)(nil)
)
// EDA is the PostgreSQL-backed EDA persistence.
@@ -14,8 +14,6 @@ import (
"github.com/samcharles93/archie-core/internal/infrastructure/postgres/postgresdb"
)
var _ storecontract.EventTypeStore = (*EDA)(nil)
func eventTypeValue(r postgresdb.EventType) (eventtype.EventType, error) {
t := eventtype.EventType{
ID: r.ID, Source: r.Source, Name: r.Name, WorkspaceID: org.WorkspaceID(r.WorkspaceID),
@@ -15,8 +15,6 @@ import (
"github.com/samcharles93/archie-core/internal/infrastructure/postgres/postgresdb"
)
var _ storecontract.HarnessSecretStore = (*EDA)(nil)
// harnessSecretPayload is the JSON shape sealed into harness_secrets.secret_enc:
// the whole token set behind one envelope, never separate plaintext columns.
type harnessSecretPayload struct {
@@ -0,0 +1,17 @@
-- +goose Up
-- A workflow.handoff ends its run and starts a linked run of the next workflow
-- on the same work item. Only the current run of an item is unique; a run
-- that handed off points at its successor and leaves the key.
ALTER TABLE tasks
ADD COLUMN handoff_from_task_id bigint REFERENCES tasks (id) ON DELETE SET NULL,
ADD COLUMN handoff_to_task_id bigint;
ALTER TABLE tasks DROP CONSTRAINT tasks_org_identity_owner_repo_number_key;
CREATE UNIQUE INDEX tasks_current_item_key ON tasks (org_id, identity, owner, repo, issue_number)
WHERE handoff_to_task_id IS NULL;
-- +goose Down
DELETE FROM tasks WHERE handoff_to_task_id IS NOT NULL;
DROP INDEX tasks_current_item_key;
ALTER TABLE tasks ADD CONSTRAINT tasks_org_identity_owner_repo_number_key
UNIQUE (org_id, identity, owner, repo, issue_number);
ALTER TABLE tasks DROP COLUMN handoff_to_task_id, DROP COLUMN handoff_from_task_id;
@@ -24,7 +24,7 @@ func (q *Queries) CallStatusDetail(ctx context.Context, taskID int64) (string, e
const enqueueCallTask = `-- name: EnqueueCallTask :one
WITH caller AS (
SELECT t.id, t.owner, t.repo, t.issue_number, t.title, t.body, t.labels, t.status, t.workflow, t.branch, t.plan, t.notes, t.pr_number, t.tokens_used, t.iterations, t.attempt, t.park_reason, t.watch_comment_id, t.park_class, t.remediation_rounds, t.retry_count, t.source, t.identity, t.binding_id, t.binding_version, t.review_payload, t.workflow_definition_version, t.workflow_definition_digest, t.workflow_definition_yaml, t.created_at, t.updated_at, t.review_cursor, t.inputs, t.org_id, t.workspace_id, t.call_parent_task_id, t.call_depth, t.outputs, t.review_gate, t.rereview_rounds, t.retry_mode, t.resume_from, t.resume_results, t.pending_reviews, t.origin, t.call_key, t.binding_signature, t.binding_addr, t.park_blocker, t.outcome_posted_status, t.plan_document FROM tasks t WHERE t.id = $1 FOR UPDATE
SELECT t.id, t.owner, t.repo, t.issue_number, t.title, t.body, t.labels, t.status, t.workflow, t.branch, t.plan, t.notes, t.pr_number, t.tokens_used, t.iterations, t.attempt, t.park_reason, t.watch_comment_id, t.park_class, t.remediation_rounds, t.retry_count, t.source, t.identity, t.binding_id, t.binding_version, t.review_payload, t.workflow_definition_version, t.workflow_definition_digest, t.workflow_definition_yaml, t.created_at, t.updated_at, t.review_cursor, t.inputs, t.org_id, t.workspace_id, t.call_parent_task_id, t.call_depth, t.outputs, t.review_gate, t.rereview_rounds, t.retry_mode, t.resume_from, t.resume_results, t.pending_reviews, t.origin, t.call_key, t.binding_signature, t.binding_addr, t.park_blocker, t.outcome_posted_status, t.plan_document, t.handoff_from_task_id, t.handoff_to_task_id FROM tasks t WHERE t.id = $1 FOR UPDATE
), number AS (
INSERT INTO chat_issue_numbers (owner, repo, last_number)
SELECT caller.owner, caller.repo, $9::bigint + 1 FROM caller
@@ -43,7 +43,7 @@ SELECT
FROM caller, number
WHERE caller.status = 'running' AND caller.call_depth + 1 <= $8::int
ON CONFLICT (call_parent_task_id, call_key) WHERE call_key <> '' AND status NOT IN ('dead', 'closed_wont_do') DO UPDATE SET call_key = EXCLUDED.call_key
RETURNING id, owner, repo, issue_number, title, body, labels, status, workflow, branch, plan, notes, pr_number, tokens_used, iterations, attempt, park_reason, watch_comment_id, park_class, remediation_rounds, retry_count, source, identity, binding_id, binding_version, review_payload, workflow_definition_version, workflow_definition_digest, workflow_definition_yaml, created_at, updated_at, review_cursor, inputs, org_id, workspace_id, call_parent_task_id, call_depth, outputs, review_gate, rereview_rounds, retry_mode, resume_from, resume_results, pending_reviews, origin, call_key, binding_signature, binding_addr, park_blocker, outcome_posted_status, plan_document
RETURNING id, owner, repo, issue_number, title, body, labels, status, workflow, branch, plan, notes, pr_number, tokens_used, iterations, attempt, park_reason, watch_comment_id, park_class, remediation_rounds, retry_count, source, identity, binding_id, binding_version, review_payload, workflow_definition_version, workflow_definition_digest, workflow_definition_yaml, created_at, updated_at, review_cursor, inputs, org_id, workspace_id, call_parent_task_id, call_depth, outputs, review_gate, rereview_rounds, retry_mode, resume_from, resume_results, pending_reviews, origin, call_key, binding_signature, binding_addr, park_blocker, outcome_posted_status, plan_document, handoff_from_task_id, handoff_to_task_id
`
type EnqueueCallTaskParams struct {
@@ -137,6 +137,8 @@ func (q *Queries) EnqueueCallTask(ctx context.Context, arg EnqueueCallTaskParams
&i.ParkBlocker,
&i.OutcomePostedStatus,
&i.PlanDocument,
&i.HandoffFromTaskID,
&i.HandoffToTaskID,
)
return i, err
}
@@ -584,6 +584,8 @@ type Task struct {
ParkBlocker string
OutcomePostedStatus string
PlanDocument string
HandoffFromTaskID pgtype.Int8
HandoffToTaskID pgtype.Int8
}
type ToolCall struct {
@@ -13,7 +13,7 @@ import (
)
const activeTasksByOrigin = `-- name: ActiveTasksByOrigin :many
SELECT id, owner, repo, issue_number, title, body, labels, status, workflow, branch, plan, notes, pr_number, tokens_used, iterations, attempt, park_reason, watch_comment_id, park_class, remediation_rounds, retry_count, source, identity, binding_id, binding_version, review_payload, workflow_definition_version, workflow_definition_digest, workflow_definition_yaml, created_at, updated_at, review_cursor, inputs, org_id, workspace_id, call_parent_task_id, call_depth, outputs, review_gate, rereview_rounds, retry_mode, resume_from, resume_results, pending_reviews, origin, call_key, binding_signature, binding_addr, park_blocker, outcome_posted_status, plan_document FROM tasks
SELECT id, owner, repo, issue_number, title, body, labels, status, workflow, branch, plan, notes, pr_number, tokens_used, iterations, attempt, park_reason, watch_comment_id, park_class, remediation_rounds, retry_count, source, identity, binding_id, binding_version, review_payload, workflow_definition_version, workflow_definition_digest, workflow_definition_yaml, created_at, updated_at, review_cursor, inputs, org_id, workspace_id, call_parent_task_id, call_depth, outputs, review_gate, rereview_rounds, retry_mode, resume_from, resume_results, pending_reviews, origin, call_key, binding_signature, binding_addr, park_blocker, outcome_posted_status, plan_document, handoff_from_task_id, handoff_to_task_id FROM tasks
WHERE origin = $1 AND origin <> '' AND status IN ('queued', 'running') AND ($2::text IS NULL OR org_id = $2)
ORDER BY id
`
@@ -85,6 +85,8 @@ func (q *Queries) ActiveTasksByOrigin(ctx context.Context, arg ActiveTasksByOrig
&i.ParkBlocker,
&i.OutcomePostedStatus,
&i.PlanDocument,
&i.HandoffFromTaskID,
&i.HandoffToTaskID,
); err != nil {
return nil, err
}
@@ -164,7 +166,7 @@ func (q *Queries) BeginRemediationTask(ctx context.Context, arg BeginRemediation
const claimByIssue = `-- name: ClaimByIssue :one
UPDATE tasks SET status = 'running', attempt = attempt + 1, outputs = '', updated_at = now()
WHERE owner = $1 AND repo = $2 AND issue_number = $3 AND status = 'queued'
RETURNING id, owner, repo, issue_number, title, body, labels, status, workflow, branch, plan, notes, pr_number, tokens_used, iterations, attempt, park_reason, watch_comment_id, park_class, remediation_rounds, retry_count, source, identity, binding_id, binding_version, review_payload, workflow_definition_version, workflow_definition_digest, workflow_definition_yaml, created_at, updated_at, review_cursor, inputs, org_id, workspace_id, call_parent_task_id, call_depth, outputs, review_gate, rereview_rounds, retry_mode, resume_from, resume_results, pending_reviews, origin, call_key, binding_signature, binding_addr, park_blocker, outcome_posted_status, plan_document
RETURNING id, owner, repo, issue_number, title, body, labels, status, workflow, branch, plan, notes, pr_number, tokens_used, iterations, attempt, park_reason, watch_comment_id, park_class, remediation_rounds, retry_count, source, identity, binding_id, binding_version, review_payload, workflow_definition_version, workflow_definition_digest, workflow_definition_yaml, created_at, updated_at, review_cursor, inputs, org_id, workspace_id, call_parent_task_id, call_depth, outputs, review_gate, rereview_rounds, retry_mode, resume_from, resume_results, pending_reviews, origin, call_key, binding_signature, binding_addr, park_blocker, outcome_posted_status, plan_document, handoff_from_task_id, handoff_to_task_id
`
type ClaimByIssueParams struct {
@@ -230,6 +232,8 @@ func (q *Queries) ClaimByIssue(ctx context.Context, arg ClaimByIssueParams) (Tas
&i.ParkBlocker,
&i.OutcomePostedStatus,
&i.PlanDocument,
&i.HandoffFromTaskID,
&i.HandoffToTaskID,
)
return i, err
}
@@ -244,7 +248,7 @@ WHERE id = (
FOR UPDATE SKIP LOCKED
LIMIT 1
)
RETURNING id, owner, repo, issue_number, title, body, labels, status, workflow, branch, plan, notes, pr_number, tokens_used, iterations, attempt, park_reason, watch_comment_id, park_class, remediation_rounds, retry_count, source, identity, binding_id, binding_version, review_payload, workflow_definition_version, workflow_definition_digest, workflow_definition_yaml, created_at, updated_at, review_cursor, inputs, org_id, workspace_id, call_parent_task_id, call_depth, outputs, review_gate, rereview_rounds, retry_mode, resume_from, resume_results, pending_reviews, origin, call_key, binding_signature, binding_addr, park_blocker, outcome_posted_status, plan_document
RETURNING id, owner, repo, issue_number, title, body, labels, status, workflow, branch, plan, notes, pr_number, tokens_used, iterations, attempt, park_reason, watch_comment_id, park_class, remediation_rounds, retry_count, source, identity, binding_id, binding_version, review_payload, workflow_definition_version, workflow_definition_digest, workflow_definition_yaml, created_at, updated_at, review_cursor, inputs, org_id, workspace_id, call_parent_task_id, call_depth, outputs, review_gate, rereview_rounds, retry_mode, resume_from, resume_results, pending_reviews, origin, call_key, binding_signature, binding_addr, park_blocker, outcome_posted_status, plan_document, handoff_from_task_id, handoff_to_task_id
`
// FOR UPDATE SKIP LOCKED replaces the SQLite single-writer assumption: two
@@ -308,6 +312,8 @@ func (q *Queries) ClaimNextTask(ctx context.Context) (Task, error) {
&i.ParkBlocker,
&i.OutcomePostedStatus,
&i.PlanDocument,
&i.HandoffFromTaskID,
&i.HandoffToTaskID,
)
return i, err
}
@@ -367,7 +373,7 @@ VALUES (
),
'default'
)
ON CONFLICT (org_id, identity, owner, repo, issue_number) DO NOTHING
ON CONFLICT (org_id, identity, owner, repo, issue_number) WHERE handoff_to_task_id IS NULL DO NOTHING
`
type EnqueueIssueParams struct {
@@ -427,7 +433,7 @@ FROM number, (SELECT COALESCE(
'org-sys'
) AS id) AS owner_org
WHERE $11::text IS NULL OR owner_org.id = $11
RETURNING id, owner, repo, issue_number, title, body, labels, status, workflow, branch, plan, notes, pr_number, tokens_used, iterations, attempt, park_reason, watch_comment_id, park_class, remediation_rounds, retry_count, source, identity, binding_id, binding_version, review_payload, workflow_definition_version, workflow_definition_digest, workflow_definition_yaml, created_at, updated_at, review_cursor, inputs, org_id, workspace_id, call_parent_task_id, call_depth, outputs, review_gate, rereview_rounds, retry_mode, resume_from, resume_results, pending_reviews, origin, call_key, binding_signature, binding_addr, park_blocker, outcome_posted_status, plan_document
RETURNING id, owner, repo, issue_number, title, body, labels, status, workflow, branch, plan, notes, pr_number, tokens_used, iterations, attempt, park_reason, watch_comment_id, park_class, remediation_rounds, retry_count, source, identity, binding_id, binding_version, review_payload, workflow_definition_version, workflow_definition_digest, workflow_definition_yaml, created_at, updated_at, review_cursor, inputs, org_id, workspace_id, call_parent_task_id, call_depth, outputs, review_gate, rereview_rounds, retry_mode, resume_from, resume_results, pending_reviews, origin, call_key, binding_signature, binding_addr, park_blocker, outcome_posted_status, plan_document, handoff_from_task_id, handoff_to_task_id
`
type InsertChatTaskParams struct {
@@ -525,6 +531,88 @@ func (q *Queries) InsertChatTask(ctx context.Context, arg InsertChatTaskParams)
&i.ParkBlocker,
&i.OutcomePostedStatus,
&i.PlanDocument,
&i.HandoffFromTaskID,
&i.HandoffToTaskID,
)
return i, err
}
const insertHandoffTask = `-- name: InsertHandoffTask :one
INSERT INTO tasks (id, owner, repo, issue_number, title, body, labels, workflow, source, identity,
org_id, workspace_id, inputs, plan, plan_document, binding_id, binding_version, binding_signature,
binding_addr, origin, call_parent_task_id, call_depth, handoff_from_task_id)
SELECT $1, t.owner, t.repo, t.issue_number, t.title, t.body, t.labels, $2, t.source, t.identity,
t.org_id, t.workspace_id, t.inputs, t.plan, t.plan_document, t.binding_id, t.binding_version, t.binding_signature,
t.binding_addr, t.origin, t.call_parent_task_id, t.call_depth, t.id
FROM tasks t WHERE t.id = $3
RETURNING id, owner, repo, issue_number, title, body, labels, status, workflow, branch, plan, notes, pr_number, tokens_used, iterations, attempt, park_reason, watch_comment_id, park_class, remediation_rounds, retry_count, source, identity, binding_id, binding_version, review_payload, workflow_definition_version, workflow_definition_digest, workflow_definition_yaml, created_at, updated_at, review_cursor, inputs, org_id, workspace_id, call_parent_task_id, call_depth, outputs, review_gate, rereview_rounds, retry_mode, resume_from, resume_results, pending_reviews, origin, call_key, binding_signature, binding_addr, park_blocker, outcome_posted_status, plan_document, handoff_from_task_id, handoff_to_task_id
`
type InsertHandoffTaskParams struct {
SuccessorID int64
Workflow string
ID int64
}
// The successor run carries the work item and the decisions made on it; its
// run state (attempts, branch, outputs, review state) starts fresh.
func (q *Queries) InsertHandoffTask(ctx context.Context, arg InsertHandoffTaskParams) (Task, error) {
row := q.db.QueryRow(ctx, insertHandoffTask, arg.SuccessorID, arg.Workflow, arg.ID)
var i Task
err := row.Scan(
&i.ID,
&i.Owner,
&i.Repo,
&i.IssueNumber,
&i.Title,
&i.Body,
&i.Labels,
&i.Status,
&i.Workflow,
&i.Branch,
&i.Plan,
&i.Notes,
&i.PrNumber,
&i.TokensUsed,
&i.Iterations,
&i.Attempt,
&i.ParkReason,
&i.WatchCommentID,
&i.ParkClass,
&i.RemediationRounds,
&i.RetryCount,
&i.Source,
&i.Identity,
&i.BindingID,
&i.BindingVersion,
&i.ReviewPayload,
&i.WorkflowDefinitionVersion,
&i.WorkflowDefinitionDigest,
&i.WorkflowDefinitionYaml,
&i.CreatedAt,
&i.UpdatedAt,
&i.ReviewCursor,
&i.Inputs,
&i.OrgID,
&i.WorkspaceID,
&i.CallParentTaskID,
&i.CallDepth,
&i.Outputs,
&i.ReviewGate,
&i.RereviewRounds,
&i.RetryMode,
&i.ResumeFrom,
&i.ResumeResults,
&i.PendingReviews,
&i.Origin,
&i.CallKey,
&i.BindingSignature,
&i.BindingAddr,
&i.ParkBlocker,
&i.OutcomePostedStatus,
&i.PlanDocument,
&i.HandoffFromTaskID,
&i.HandoffToTaskID,
)
return i, err
}
@@ -609,7 +697,8 @@ const listTaskSummaries = `-- name: ListTaskSummaries :many
SELECT id, owner, repo, issue_number, title, status, workflow,
pr_number, tokens_used, iterations, attempt, park_reason, retry_count,
created_at, updated_at, plan, plan_document, source, identity, binding_id, binding_version,
outputs, review_gate, rereview_rounds, org_id, park_class, park_blocker
outputs, review_gate, rereview_rounds, org_id, park_class, park_blocker,
handoff_from_task_id, handoff_to_task_id
FROM tasks
WHERE ($1::text IS NULL OR org_id = $1)
AND (cardinality($2::text[]) = 0 OR status = ANY($2::text[]))
@@ -630,33 +719,35 @@ type ListTaskSummariesParams struct {
}
type ListTaskSummariesRow struct {
ID int64
Owner string
Repo string
IssueNumber int64
Title string
Status string
Workflow string
PrNumber int64
TokensUsed int64
Iterations int64
Attempt int64
ParkReason string
RetryCount int64
CreatedAt time.Time
UpdatedAt time.Time
Plan string
PlanDocument string
Source string
Identity string
BindingID string
BindingVersion int64
Outputs string
ReviewGate string
RereviewRounds int64
OrgID string
ParkClass string
ParkBlocker string
ID int64
Owner string
Repo string
IssueNumber int64
Title string
Status string
Workflow string
PrNumber int64
TokensUsed int64
Iterations int64
Attempt int64
ParkReason string
RetryCount int64
CreatedAt time.Time
UpdatedAt time.Time
Plan string
PlanDocument string
Source string
Identity string
BindingID string
BindingVersion int64
Outputs string
ReviewGate string
RereviewRounds int64
OrgID string
ParkClass string
ParkBlocker string
HandoffFromTaskID pgtype.Int8
HandoffToTaskID pgtype.Int8
}
// The dashboard's list. This projection is deliberately narrow: Plan gates
@@ -712,6 +803,8 @@ func (q *Queries) ListTaskSummaries(ctx context.Context, arg ListTaskSummariesPa
&i.OrgID,
&i.ParkClass,
&i.ParkBlocker,
&i.HandoffFromTaskID,
&i.HandoffToTaskID,
); err != nil {
return nil, err
}
@@ -763,6 +856,25 @@ func (q *Queries) LockTaskStatus(ctx context.Context, id int64) (string, error)
return status, err
}
const markHandedOff = `-- name: MarkHandedOff :execrows
UPDATE tasks SET handoff_to_task_id = $1, updated_at = now()
WHERE id = $2 AND handoff_to_task_id IS NULL
`
type MarkHandedOffParams struct {
SuccessorID pgtype.Int8
ID int64
}
// The run leaves the item's unique key before its successor takes it.
func (q *Queries) MarkHandedOff(ctx context.Context, arg MarkHandedOffParams) (int64, error) {
result, err := q.db.Exec(ctx, markHandedOff, arg.SuccessorID, arg.ID)
if err != nil {
return 0, err
}
return result.RowsAffected(), nil
}
const markOutcomePosted = `-- name: MarkOutcomePosted :exec
UPDATE tasks SET outcome_posted_status = COALESCE(NULLIF($1::text, ''), status) WHERE id = $2
`
@@ -780,6 +892,17 @@ func (q *Queries) MarkOutcomePosted(ctx context.Context, arg MarkOutcomePostedPa
return err
}
const nextTaskID = `-- name: NextTaskID :one
SELECT nextval(pg_get_serial_sequence('tasks', 'id'))::bigint
`
func (q *Queries) NextTaskID(ctx context.Context) (int64, error) {
row := q.db.QueryRow(ctx, nextTaskID)
var column_1 int64
err := row.Scan(&column_1)
return column_1, err
}
const parkTask = `-- name: ParkTask :execrows
UPDATE tasks
SET status = 'parked', park_reason = $2, park_class = $3, park_blocker = $5, updated_at = now()
@@ -1014,7 +1137,7 @@ func (q *Queries) StampTaskBinding(ctx context.Context, arg StampTaskBindingPara
}
const taskByID = `-- name: TaskByID :one
SELECT id, owner, repo, issue_number, title, body, labels, status, workflow, branch, plan, notes, pr_number, tokens_used, iterations, attempt, park_reason, watch_comment_id, park_class, remediation_rounds, retry_count, source, identity, binding_id, binding_version, review_payload, workflow_definition_version, workflow_definition_digest, workflow_definition_yaml, created_at, updated_at, review_cursor, inputs, org_id, workspace_id, call_parent_task_id, call_depth, outputs, review_gate, rereview_rounds, retry_mode, resume_from, resume_results, pending_reviews, origin, call_key, binding_signature, binding_addr, park_blocker, outcome_posted_status, plan_document FROM tasks WHERE id = $1 AND ($2::text IS NULL OR org_id = $2)
SELECT id, owner, repo, issue_number, title, body, labels, status, workflow, branch, plan, notes, pr_number, tokens_used, iterations, attempt, park_reason, watch_comment_id, park_class, remediation_rounds, retry_count, source, identity, binding_id, binding_version, review_payload, workflow_definition_version, workflow_definition_digest, workflow_definition_yaml, created_at, updated_at, review_cursor, inputs, org_id, workspace_id, call_parent_task_id, call_depth, outputs, review_gate, rereview_rounds, retry_mode, resume_from, resume_results, pending_reviews, origin, call_key, binding_signature, binding_addr, park_blocker, outcome_posted_status, plan_document, handoff_from_task_id, handoff_to_task_id FROM tasks WHERE id = $1 AND ($2::text IS NULL OR org_id = $2)
`
type TaskByIDParams struct {
@@ -1077,12 +1200,14 @@ func (q *Queries) TaskByID(ctx context.Context, arg TaskByIDParams) (Task, error
&i.ParkBlocker,
&i.OutcomePostedStatus,
&i.PlanDocument,
&i.HandoffFromTaskID,
&i.HandoffToTaskID,
)
return i, err
}
const taskByIssue = `-- name: TaskByIssue :one
SELECT id, owner, repo, issue_number, title, body, labels, status, workflow, branch, plan, notes, pr_number, tokens_used, iterations, attempt, park_reason, watch_comment_id, park_class, remediation_rounds, retry_count, source, identity, binding_id, binding_version, review_payload, workflow_definition_version, workflow_definition_digest, workflow_definition_yaml, created_at, updated_at, review_cursor, inputs, org_id, workspace_id, call_parent_task_id, call_depth, outputs, review_gate, rereview_rounds, retry_mode, resume_from, resume_results, pending_reviews, origin, call_key, binding_signature, binding_addr, park_blocker, outcome_posted_status, plan_document FROM tasks WHERE owner = $1 AND repo = $2 AND issue_number = $3
SELECT id, owner, repo, issue_number, title, body, labels, status, workflow, branch, plan, notes, pr_number, tokens_used, iterations, attempt, park_reason, watch_comment_id, park_class, remediation_rounds, retry_count, source, identity, binding_id, binding_version, review_payload, workflow_definition_version, workflow_definition_digest, workflow_definition_yaml, created_at, updated_at, review_cursor, inputs, org_id, workspace_id, call_parent_task_id, call_depth, outputs, review_gate, rereview_rounds, retry_mode, resume_from, resume_results, pending_reviews, origin, call_key, binding_signature, binding_addr, park_blocker, outcome_posted_status, plan_document, handoff_from_task_id, handoff_to_task_id FROM tasks WHERE owner = $1 AND repo = $2 AND issue_number = $3 AND handoff_to_task_id IS NULL
AND ($4::text IS NULL OR org_id = $4)
`
@@ -1153,12 +1278,14 @@ func (q *Queries) TaskByIssue(ctx context.Context, arg TaskByIssueParams) (Task,
&i.ParkBlocker,
&i.OutcomePostedStatus,
&i.PlanDocument,
&i.HandoffFromTaskID,
&i.HandoffToTaskID,
)
return i, err
}
const taskByPR = `-- name: TaskByPR :one
SELECT id, owner, repo, issue_number, title, body, labels, status, workflow, branch, plan, notes, pr_number, tokens_used, iterations, attempt, park_reason, watch_comment_id, park_class, remediation_rounds, retry_count, source, identity, binding_id, binding_version, review_payload, workflow_definition_version, workflow_definition_digest, workflow_definition_yaml, created_at, updated_at, review_cursor, inputs, org_id, workspace_id, call_parent_task_id, call_depth, outputs, review_gate, rereview_rounds, retry_mode, resume_from, resume_results, pending_reviews, origin, call_key, binding_signature, binding_addr, park_blocker, outcome_posted_status, plan_document FROM tasks WHERE owner = $1 AND repo = $2 AND pr_number = $3 AND status = $4
SELECT id, owner, repo, issue_number, title, body, labels, status, workflow, branch, plan, notes, pr_number, tokens_used, iterations, attempt, park_reason, watch_comment_id, park_class, remediation_rounds, retry_count, source, identity, binding_id, binding_version, review_payload, workflow_definition_version, workflow_definition_digest, workflow_definition_yaml, created_at, updated_at, review_cursor, inputs, org_id, workspace_id, call_parent_task_id, call_depth, outputs, review_gate, rereview_rounds, retry_mode, resume_from, resume_results, pending_reviews, origin, call_key, binding_signature, binding_addr, park_blocker, outcome_posted_status, plan_document, handoff_from_task_id, handoff_to_task_id FROM tasks WHERE owner = $1 AND repo = $2 AND pr_number = $3 AND status = $4
AND ($5::text IS NULL OR org_id = $5)
`
@@ -1234,6 +1361,8 @@ func (q *Queries) TaskByPR(ctx context.Context, arg TaskByPRParams) (Task, error
&i.ParkBlocker,
&i.OutcomePostedStatus,
&i.PlanDocument,
&i.HandoffFromTaskID,
&i.HandoffToTaskID,
)
return i, err
}
@@ -1273,9 +1402,10 @@ func (q *Queries) TransitionTask(ctx context.Context, arg TransitionTaskParams)
}
const unpostedOutcomes = `-- name: UnpostedOutcomes :many
SELECT id, owner, repo, issue_number, title, body, labels, status, workflow, branch, plan, notes, pr_number, tokens_used, iterations, attempt, park_reason, watch_comment_id, park_class, remediation_rounds, retry_count, source, identity, binding_id, binding_version, review_payload, workflow_definition_version, workflow_definition_digest, workflow_definition_yaml, created_at, updated_at, review_cursor, inputs, org_id, workspace_id, call_parent_task_id, call_depth, outputs, review_gate, rereview_rounds, retry_mode, resume_from, resume_results, pending_reviews, origin, call_key, binding_signature, binding_addr, park_blocker, outcome_posted_status, plan_document FROM tasks
SELECT id, owner, repo, issue_number, title, body, labels, status, workflow, branch, plan, notes, pr_number, tokens_used, iterations, attempt, park_reason, watch_comment_id, park_class, remediation_rounds, retry_count, source, identity, binding_id, binding_version, review_payload, workflow_definition_version, workflow_definition_digest, workflow_definition_yaml, created_at, updated_at, review_cursor, inputs, org_id, workspace_id, call_parent_task_id, call_depth, outputs, review_gate, rereview_rounds, retry_mode, resume_from, resume_results, pending_reviews, origin, call_key, binding_signature, binding_addr, park_blocker, outcome_posted_status, plan_document, handoff_from_task_id, handoff_to_task_id FROM tasks
WHERE origin <> '' AND outcome_posted_status <> status
AND status IN ('merged', 'rejected', 'dead', 'closed_wont_do', 'completed', 'waiting_approval', 'pr_open')
AND handoff_to_task_id IS NULL
AND ($1::text IS NULL OR org_id = $1)
ORDER BY id
LIMIT 50
@@ -1344,6 +1474,8 @@ func (q *Queries) UnpostedOutcomes(ctx context.Context, scopeOrg pgtype.Text) ([
&i.ParkBlocker,
&i.OutcomePostedStatus,
&i.PlanDocument,
&i.HandoffFromTaskID,
&i.HandoffToTaskID,
); err != nil {
return nil, err
}
@@ -20,7 +20,7 @@ RETURNING *;
SELECT * FROM tasks WHERE id = sqlc.arg(id) AND (sqlc.narg(scope_org)::text IS NULL OR org_id = sqlc.narg(scope_org));
-- name: TaskByIssue :one
SELECT * FROM tasks WHERE owner = sqlc.arg(owner) AND repo = sqlc.arg(repo) AND issue_number = sqlc.arg(issue_number)
SELECT * FROM tasks WHERE owner = sqlc.arg(owner) AND repo = sqlc.arg(repo) AND issue_number = sqlc.arg(issue_number) AND handoff_to_task_id IS NULL
AND (sqlc.narg(scope_org)::text IS NULL OR org_id = sqlc.narg(scope_org));
-- name: TaskByPR :one
@@ -43,7 +43,8 @@ SELECT * FROM tasks WHERE owner = sqlc.arg(owner) AND repo = sqlc.arg(repo) AND
SELECT id, owner, repo, issue_number, title, status, workflow,
pr_number, tokens_used, iterations, attempt, park_reason, retry_count,
created_at, updated_at, plan, plan_document, source, identity, binding_id, binding_version,
outputs, review_gate, rereview_rounds, org_id, park_class, park_blocker
outputs, review_gate, rereview_rounds, org_id, park_class, park_blocker,
handoff_from_task_id, handoff_to_task_id
FROM tasks
WHERE (sqlc.narg(scope_org)::text IS NULL OR org_id = sqlc.narg(scope_org))
AND (cardinality(@statuses::text[]) = 0 OR status = ANY(@statuses::text[]))
@@ -109,6 +110,7 @@ ORDER BY id;
SELECT * FROM tasks
WHERE origin <> '' AND outcome_posted_status <> status
AND status IN ('merged', 'rejected', 'dead', 'closed_wont_do', 'completed', 'waiting_approval', 'pr_open')
AND handoff_to_task_id IS NULL
AND (sqlc.narg(scope_org)::text IS NULL OR org_id = sqlc.narg(scope_org))
ORDER BY id
LIMIT 50;
@@ -199,7 +201,7 @@ VALUES (
),
'default'
)
ON CONFLICT (org_id, identity, owner, repo, issue_number) DO NOTHING;
ON CONFLICT (org_id, identity, owner, repo, issue_number) WHERE handoff_to_task_id IS NULL DO NOTHING;
-- name: ClaimByIssue :one
-- The claim starts the attempt's output set empty, as ClaimNextTask does:
@@ -291,3 +293,23 @@ DELETE FROM tasks WHERE id = $1 AND status = $2;
-- name: ListOpenPRs :many
SELECT id, owner, repo, issue_number, pr_number, status, source, identity, attempt, review_cursor, watch_comment_id, outputs
FROM tasks WHERE status = 'pr_open' AND (sqlc.narg(scope_org)::text IS NULL OR org_id = sqlc.narg(scope_org));
-- name: NextTaskID :one
SELECT nextval(pg_get_serial_sequence('tasks', 'id'))::bigint;
-- name: MarkHandedOff :execrows
-- The run leaves the item's unique key before its successor takes it.
UPDATE tasks SET handoff_to_task_id = sqlc.arg(successor_id), updated_at = now()
WHERE id = sqlc.arg(id) AND handoff_to_task_id IS NULL;
-- name: InsertHandoffTask :one
-- The successor run carries the work item and the decisions made on it; its
-- run state (attempts, branch, outputs, review state) starts fresh.
INSERT INTO tasks (id, owner, repo, issue_number, title, body, labels, workflow, source, identity,
org_id, workspace_id, inputs, plan, plan_document, binding_id, binding_version, binding_signature,
binding_addr, origin, call_parent_task_id, call_depth, handoff_from_task_id)
SELECT sqlc.arg(successor_id), t.owner, t.repo, t.issue_number, t.title, t.body, t.labels, sqlc.arg(workflow), t.source, t.identity,
t.org_id, t.workspace_id, t.inputs, t.plan, t.plan_document, t.binding_id, t.binding_version, t.binding_signature,
t.binding_addr, t.origin, t.call_parent_task_id, t.call_depth, t.id
FROM tasks t WHERE t.id = sqlc.arg(id)
RETURNING *;
@@ -11,8 +11,6 @@ import (
"github.com/samcharles93/archie-core/internal/infrastructure/postgres/postgresdb"
)
var _ storecontract.RunCredentialStore = (*Store)(nil)
func (s *Store) PutRunCredential(ctx context.Context, digest [32]byte, taskID int64, expires time.Time) error {
return s.queries().InsertRunCredential(ctx, postgresdb.InsertRunCredentialParams{Digest: digest[:], TaskID: taskID, ExpiresAt: expires})
}
@@ -13,8 +13,6 @@ import (
"github.com/samcharles93/archie-core/internal/infrastructure/postgres/postgresdb"
)
var _ storecontract.SourceStore = (*EDA)(nil)
// InsertSource stores a new source in one of the acting org's workspaces,
// its default one when src names none; a workspace the org does not have is
// ErrInvalidWorkspace. The path is the primary key, so a taken path is
+66
View File
@@ -124,6 +124,8 @@ func taskFromRow(t postgresdb.Task) *workflow.Task {
ParkBlocker: t.ParkBlocker,
RemediationRounds: int(t.RemediationRounds),
CallParentTaskID: t.CallParentTaskID,
HandoffFromTaskID: t.HandoffFromTaskID.Int64,
HandoffToTaskID: t.HandoffToTaskID.Int64,
CallDepth: int(t.CallDepth),
Origin: t.Origin,
CreatedAt: t.CreatedAt,
@@ -134,9 +136,20 @@ func taskFromRow(t postgresdb.Task) *workflow.Task {
// Compile-time checks: *Store satisfies every surface the SQLite store did.
var (
_ storecontract.TaskStore = (*Store)(nil)
_ storecontract.TaskLifecycle = (*Store)(nil)
_ storecontract.TaskParker = (*Store)(nil)
_ storecontract.TaskQueries = (*Store)(nil)
_ storecontract.TaskEvents = (*Store)(nil)
_ storecontract.TaskArchiver = (*Store)(nil)
_ storecontract.TaskRetryer = (*Store)(nil)
_ storecontract.ReviewGateResponder = (*Store)(nil)
_ storecontract.RemediationStarter = (*Store)(nil)
_ storecontract.OutcomePoster = (*Store)(nil)
_ storecontract.StepRecorder = (*Store)(nil)
_ storecontract.StepReader = (*Store)(nil)
_ storecontract.ExecutionCanceller = (*Store)(nil)
_ storecontract.BindingTaskCreator = (*Store)(nil)
_ storecontract.RunCredentialStore = (*Store)(nil)
_ storecontract.ConfigSnapshotStore = (*Store)(nil)
_ storecontract.ChannelStatusStore = (*Store)(nil)
_ storecontract.ApplyStatusStore = (*Store)(nil)
@@ -779,6 +792,7 @@ func (s *Store) TasksPage(ctx context.Context, page storecontract.TaskPage) ([]w
Outputs: outputs,
ReviewGate: r.ReviewGate, RereviewRounds: int(r.RereviewRounds),
Org: org.OrgID(r.OrgID), ParkClass: r.ParkClass, ParkBlocker: r.ParkBlocker,
HandoffFromTaskID: r.HandoffFromTaskID.Int64, HandoffToTaskID: r.HandoffToTaskID.Int64,
})
}
return tasks, nil
@@ -875,6 +889,58 @@ func (s *Store) StartCall(ctx context.Context, callerTaskID int64, callKey, wf s
return taskFromRow(callee), nil
}
// Handoff completes the running task fromTaskID and starts its linked
// successor on the same work item under wf, in one transaction: the item is
// never without a current run, and never has two.
func (s *Store) Handoff(ctx context.Context, fromTaskID int64, wf, detail string) (*workflow.Task, error) {
tx, err := s.pool.Begin(ctx)
if err != nil {
return nil, err
}
defer func() { _ = tx.Rollback(ctx) }()
q := postgresdb.New(tx)
from, to := workflow.StatusRunning, workflow.StatusCompleted
if err := guardTransition(ctx, q, fromTaskID, from, to); err != nil {
return nil, err
}
n, err := q.TransitionTask(ctx, postgresdb.TransitionTaskParams{
ID: fromTaskID, Status: to, ParkClass: taskstate.ParkNeedsDecision, Status_2: from,
})
if err != nil {
return nil, err
}
if n == 0 {
return nil, storecontract.ErrStaleTransition
}
if err := recordTransition(ctx, q, postgresdb.InsertTransitionParams{
TaskID: fromTaskID, FromStatus: from, ToStatus: to, Detail: clip(detail, 4000),
}); err != nil {
return nil, err
}
successor, err := q.NextTaskID(ctx)
if err != nil {
return nil, err
}
if n, err := q.MarkHandedOff(ctx, postgresdb.MarkHandedOffParams{ID: fromTaskID, SuccessorID: pgtype.Int8{Int64: successor, Valid: true}}); err != nil {
return nil, err
} else if n == 0 {
return nil, fmt.Errorf("store: task %d already handed off: %w", fromTaskID, storecontract.ErrStaleTransition)
}
row, err := q.InsertHandoffTask(ctx, postgresdb.InsertHandoffTaskParams{ID: fromTaskID, SuccessorID: successor, Workflow: wf})
if err != nil {
return nil, err
}
if err := recordTransition(ctx, q, postgresdb.InsertTransitionParams{
TaskID: row.ID, FromStatus: "", ToStatus: workflow.StatusQueued, Detail: fmt.Sprintf("handed off from task %d", fromTaskID),
}); err != nil {
return nil, err
}
if err := tx.Commit(ctx); err != nil {
return nil, err
}
return taskFromRow(row), nil
}
// callerIdentity is the identity the caller's pinned workflow declares, the
// one its callees are started by. The pin does not change while the caller's
// call step runs.
@@ -111,6 +111,18 @@ var (
_ storecontract.PresenceStore = (*Client)(nil)
_ storecontract.CaptureRefusalStore = (*Client)(nil)
_ storecontract.ExecutionCanceller = (*Client)(nil)
_ storecontract.TaskLifecycle = (*Client)(nil)
_ storecontract.TaskParker = (*Client)(nil)
_ storecontract.TaskQueries = (*Client)(nil)
_ storecontract.TaskEvents = (*Client)(nil)
_ storecontract.TaskArchiver = (*Client)(nil)
_ storecontract.TaskRetryer = (*Client)(nil)
_ storecontract.ReviewGateResponder = (*Client)(nil)
_ storecontract.RemediationStarter = (*Client)(nil)
_ storecontract.OutcomePoster = (*Client)(nil)
_ storecontract.StepReader = (*Client)(nil)
_ storecontract.DispatchLedger = (*Client)(nil)
_ storecontract.EventTypeStore = (*Client)(nil)
_ identity.Repository = (*Client)(nil)
_ storepkg.Installations = (*Client)(nil)
_ identity.PersonalTokens = (*Client)(nil)
@@ -90,6 +90,40 @@ func (s *accountServer) RevokePersonalToken(ctx context.Context, r *pb.RevokePer
return &pb.RevokePersonalTokenResponse{}, nil
}
// BindIdentityPersonalToken binds a token to a named non-user identity, the
// credential a service account acts with. The audit actor is the caller's
// principal, so the request never names who did it.
func (s *accountServer) BindIdentityPersonalToken(ctx context.Context, r *pb.BindIdentityPersonalTokenRequest) (*pb.BindIdentityPersonalTokenResponse, error) {
if s.deps.PersonalTokens == nil {
return nil, status.Error(codes.Unavailable, "personal tokens unavailable")
}
if s.deps.Identities == nil {
return nil, status.Error(codes.Unavailable, "identity store unavailable")
}
caller, ok := access.PrincipalFromContext(ctx)
if !ok || caller.IdentityID == "" {
return nil, status.Error(codes.Unauthenticated, "binding an identity token requires a principal")
}
target, err := s.deps.Identities.Get(ctx, identity.IdentityID(r.GetIdentityId()))
if err != nil {
return nil, s.identityStatus(ctx, err)
}
if target.Kind == identity.KindUser {
return nil, s.identityStatus(ctx, fmt.Errorf("%w: personal tokens belong to a signed-in person", identity.ErrInvalid))
}
if !target.CanAct() {
return nil, s.identityStatus(ctx, fmt.Errorf("%w: identity is not active", identity.ErrIllegalTransition))
}
subject, err := tokenSubject(r.GetId())
if err != nil {
return nil, err
}
if err := s.deps.PersonalTokens.BindPersonalToken(ctx, target.ID, subject, r.GetName(), optionalTime(r.GetExpiresAt()), tokenAudit(ctx, caller.IdentityID, "bind", r.GetId())); err != nil {
return nil, s.identityStatus(ctx, err)
}
return &pb.BindIdentityPersonalTokenResponse{}, nil
}
func (c *Client) AddPersonalToken(ctx context.Context, subject identity.Subject, name string, expiresAt *time.Time) error {
_, err := c.account.AddPersonalToken(ctx, &pb.AddPersonalTokenRequest{Id: subject.Subject, Name: name, ExpiresAt: optionalTimestamp(expiresAt)})
return identityClientError(err)
@@ -115,6 +149,15 @@ func (c *Client) RevokePersonalToken(ctx context.Context, id string) error {
return identityClientError(err)
}
// BindPersonalToken binds a token to a named identity rather than the caller's
// principal. The audit argument is accepted for the caller-side contract but
// never sent: the store attributes the bind to the caller's own principal, so
// a request cannot claim an actor.
func (c *Client) BindPersonalToken(ctx context.Context, id identity.IdentityID, subject identity.Subject, name string, expiresAt *time.Time, _ identity.Audit) error {
_, err := c.account.BindIdentityPersonalToken(ctx, &pb.BindIdentityPersonalTokenRequest{IdentityId: string(id), Id: subject.Subject, Name: name, ExpiresAt: optionalTimestamp(expiresAt)})
return identityClientError(err)
}
func optionalTimestamp(t *time.Time) *timestamppb.Timestamp {
if t == nil {
return nil
@@ -6,6 +6,10 @@ import (
"testing"
"time"
"google.golang.org/grpc/codes"
"google.golang.org/grpc/status"
"google.golang.org/protobuf/types/known/timestamppb"
pb "github.com/samcharles93/archie-core/internal/contracts/state/v1"
"github.com/samcharles93/archie-core/internal/domain/access"
"github.com/samcharles93/archie-core/internal/domain/identity"
@@ -49,3 +53,88 @@ func TestPersonalTokenOwner(t *testing.T) {
})
}
}
type bindCalls struct {
identity.PersonalTokenStore
bound bool
id identity.IdentityID
subject identity.Subject
name string
expires *time.Time
audit identity.Audit
}
func (b *bindCalls) BindPersonalToken(_ context.Context, id identity.IdentityID, subject identity.Subject, name string, expiresAt *time.Time, audit identity.Audit) error {
b.bound, b.id, b.subject, b.name, b.expires, b.audit = true, id, subject, name, expiresAt, audit
return nil
}
type bindIdentityRepo struct {
identity.Repository
values map[identity.IdentityID]identity.Identity
}
func (r bindIdentityRepo) Get(_ context.Context, id identity.IdentityID) (identity.Identity, error) {
if value, ok := r.values[id]; ok {
return value, nil
}
return identity.Identity{}, identity.ErrNotFound
}
// TestBindIdentityPersonalToken pins the named-identity bind: a service
// account gets the token with the caller as audit actor, while a user target,
// an unknown or inactive identity, a malformed token id and a caller with no
// principal are all refused without binding.
func TestBindIdentityPersonalToken(t *testing.T) {
id := strings.Repeat("ab", 32)
sa := identity.Identity{ID: "sa1", Kind: identity.KindServiceAccount, DisplayName: "Canary", Lifecycle: identity.LifecycleActive}
user := identity.Identity{ID: "u1", Kind: identity.KindUser, DisplayName: "Alice", Lifecycle: identity.LifecycleActive}
suspended := identity.Identity{ID: "sa2", Kind: identity.KindBot, DisplayName: "Old", Lifecycle: identity.LifecycleSuspended}
repo := bindIdentityRepo{values: map[identity.IdentityID]identity.Identity{sa.ID: sa, user.ID: user, suspended.ID: suspended}}
tests := []struct {
name string
target identity.IdentityID
tokenID string
noPrincipal bool
wantCode codes.Code
wantBound bool
}{
{"service account binds with the caller as actor", sa.ID, id, false, codes.OK, true},
{"user target is refused", user.ID, id, false, codes.FailedPrecondition, false},
{"unknown identity is refused", "nope", id, false, codes.NotFound, false},
{"inactive identity is refused", suspended.ID, id, false, codes.FailedPrecondition, false},
{"malformed token id is refused", sa.ID, "not-hex", false, codes.FailedPrecondition, false},
{"no principal is refused", sa.ID, id, true, codes.Unauthenticated, false},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
store := &bindCalls{}
s := &accountServer{server: &server{deps: Deps{PersonalTokens: store, Identities: repo}}}
ctx := context.Background()
if !tt.noPrincipal {
ctx = access.WithPrincipal(ctx, access.Principal{IdentityID: "alice"})
}
expires := time.Now().UTC().Add(30 * 24 * time.Hour).Truncate(time.Second)
_, err := s.BindIdentityPersonalToken(ctx, &pb.BindIdentityPersonalTokenRequest{IdentityId: string(tt.target), Id: tt.tokenID, Name: "ci", ExpiresAt: timestamppb.New(expires)})
if got := status.Code(err); got != tt.wantCode {
t.Fatalf("code=%v want=%v err=%v", got, tt.wantCode, err)
}
if store.bound != tt.wantBound {
t.Fatalf("bound=%v want=%v", store.bound, tt.wantBound)
}
if !tt.wantBound {
return
}
wantSubject := identity.Subject{Issuer: identity.PersonalTokenIssuer, Subject: id}
if store.id != sa.ID || store.name != "ci" || store.subject != wantSubject {
t.Fatalf("bind id=%q name=%q subject=%+v, want %q %q %+v", store.id, store.name, store.subject, sa.ID, "ci", wantSubject)
}
if store.expires == nil || !store.expires.Equal(expires) {
t.Fatalf("expires=%v want %v", store.expires, expires)
}
if store.audit.ActorID != "alice" || store.audit.RequestID != "personal-token:bind:"+id {
t.Fatalf("audit=%+v, want actor alice request personal-token:bind:<id>", store.audit)
}
})
}
}
@@ -149,6 +149,13 @@ var taskScopedTargets = map[string]func(any) int64{
}
return r.CallerTaskId
},
pb.TaskService_HandoffTask_FullMethodName: func(req any) int64 {
r, ok := req.(*pb.HandoffTaskRequest)
if !ok {
return 0
}
return r.TaskId
},
pb.TaskService_WorkflowCallStatus_FullMethodName: func(req any) int64 {
r, ok := req.(*pb.WorkflowCallStatusRequest)
if !ok {
+22
View File
@@ -267,6 +267,20 @@ func (s *taskServer) EnqueueCallTask(ctx context.Context, r *pb.EnqueueCallTaskR
return &pb.EnqueueCallTaskResponse{Task: taskProto(t), Accepted: true}, nil
}
func (s *taskServer) HandoffTask(ctx context.Context, r *pb.HandoffTaskRequest) (*pb.HandoffTaskResponse, error) {
if s.deps.WorkflowCalls == nil {
return nil, errWorkflowCallerUnavailable
}
if r.TaskId <= 0 || r.Workflow == "" {
return nil, status.Error(codes.InvalidArgument, "task id and workflow are required")
}
t, err := s.deps.WorkflowCalls.Handoff(ctx, r.TaskId, r.Workflow, r.Detail)
if err != nil {
return nil, s.logErr("HandoffTask", err)
}
return &pb.HandoffTaskResponse{Task: taskProto(t)}, nil
}
func (s *taskServer) WorkflowCallStatus(ctx context.Context, r *pb.WorkflowCallStatusRequest) (*pb.WorkflowCallStatusResponse, error) {
if s.deps.WorkflowCalls == nil {
return nil, errWorkflowCallerUnavailable
@@ -485,6 +499,14 @@ func (c *Client) StartCall(ctx context.Context, callerTaskID int64, callKey, wf
return taskValue(r.Task), nil
}
func (c *Client) Handoff(ctx context.Context, fromTaskID int64, wf, detail string) (*task.Task, error) {
r, err := c.task.HandoffTask(ctx, &pb.HandoffTaskRequest{TaskId: fromTaskID, Workflow: wf, Detail: detail})
if err != nil {
return nil, unmapError(err)
}
return taskValue(r.Task), nil
}
func (c *Client) CallStatus(ctx context.Context, callerTaskID, callTaskID int64) (string, string, map[string]any, error) {
r, err := c.task.WorkflowCallStatus(ctx, &pb.WorkflowCallStatusRequest{CallerTaskId: callerTaskID, CallTaskId: callTaskID})
if err != nil {
@@ -85,6 +85,8 @@ func taskProto(t *task.Task) *pb.Task { // mirror-image field-by-field proto<->d
Org: string(t.Org),
Workspace: string(t.Workspace),
CallParentTaskId: t.CallParentTaskID,
HandoffFromTaskId: t.HandoffFromTaskID,
HandoffToTaskId: t.HandoffToTaskID,
CallDepth: int32(t.CallDepth),
Origin: t.Origin,
BindingSignature: t.BindingSignature,
@@ -122,6 +124,8 @@ func taskValue(t *pb.Task) *task.Task { // see taskProto above
Org: org.OrgID(t.Org),
Workspace: org.WorkspaceID(t.Workspace),
CallParentTaskID: t.CallParentTaskId,
HandoffFromTaskID: t.HandoffFromTaskId,
HandoffToTaskID: t.HandoffToTaskId,
CallDepth: int(t.CallDepth),
Origin: t.Origin,
BindingSignature: t.BindingSignature,
+37 -23
View File
@@ -6,14 +6,17 @@ import (
)
// The canary host installs each release from the next channel. Its behaviour
// in the 24 hours after a release is compared with its own behaviour in the
// 24 hours before, because there's only one production host to compare with.
// in the window after a release was installed is compared with its own
// behaviour in the window before, because there's only one production host to
// compare with.
const (
// CanaryWindow is how long a release runs on the canary before it can be
// promoted. A day covers one full cycle of schedules and traffic.
// CanaryWindow is the default window a release runs on the canary before
// it can be promoted. A day covers one full cycle of schedules and
// traffic.
CanaryWindow = 24 * time.Hour
// canarySettle is how long the install and restart get, counted from when
// the release was issued, before any check can fail the release.
// the release was installed on the canary, before any check can fail the
// release.
canarySettle = 30 * time.Minute
// minFinished is the fewest finished tasks needed to judge the park rate.
minFinished = 5
@@ -47,7 +50,7 @@ type CanarySignals struct {
Installed string
Ready bool
// NewActionAlerts are open action-severity alerts first raised since
// the release was issued.
// the release was installed.
NewActionAlerts []string
Baseline CanaryWindowSignals
Canary CanaryWindowSignals
@@ -59,11 +62,20 @@ type CanaryVerdict struct {
Reasons []string
}
// JudgeCanary decides what to do with a release, given how long ago it was
// issued. Once the release has settled, a failed hard check rolls it back at
// any time. Error and park rates are compared with the baseline only after the
// full window has passed.
func JudgeCanary(version string, elapsed time.Duration, s CanarySignals) CanaryVerdict {
// JudgeCanaryNow decides from the current signals alone, for an operator
// promoting without waiting out the window: the settle and window waits are
// skipped by construction, every hard and relative check still applies.
func JudgeCanaryNow(version string, now time.Time, window time.Duration, s CanarySignals) CanaryVerdict {
return JudgeCanary(version, now.Add(-(window + canarySettle)), now, window, s)
}
// JudgeCanary decides what to do with a release, given when it was installed
// on the canary host. Once the release has settled, a failed hard check rolls
// it back at any time. Error and park rates are judged against the baseline
// whenever both windows carry enough data, so a breach rolls the release back
// even mid-window; a full window with no failure promotes it.
func JudgeCanary(version string, installedAt, now time.Time, window time.Duration, s CanarySignals) CanaryVerdict {
elapsed := now.Sub(installedAt)
if elapsed < canarySettle {
return CanaryVerdict{Decision: CanaryWait, Reasons: []string{"settling after install"}}
}
@@ -80,11 +92,8 @@ func JudgeCanary(version string, elapsed time.Duration, s CanarySignals) CanaryV
if len(failed) > 0 {
return CanaryVerdict{Decision: CanaryRollBack, Reasons: failed}
}
if elapsed < CanaryWindow {
return CanaryVerdict{Decision: CanaryWait, Reasons: []string{fmt.Sprintf("%s of %s elapsed", elapsed.Round(time.Minute), CanaryWindow)}}
}
var notes []string
if reason, ok := judgeErrors(s.Baseline, s.Canary); !ok {
if reason, ok := judgeErrors(s.Baseline, s.Canary, window); !ok {
failed = append(failed, reason)
} else {
notes = append(notes, reason)
@@ -97,32 +106,37 @@ func JudgeCanary(version string, elapsed time.Duration, s CanarySignals) CanaryV
if len(failed) > 0 {
return CanaryVerdict{Decision: CanaryRollBack, Reasons: failed}
}
if elapsed < window {
reasons := append([]string{fmt.Sprintf("%s of %s elapsed since the install", elapsed.Round(time.Minute), window)}, notes...)
return CanaryVerdict{Decision: CanaryWait, Reasons: reasons}
}
return CanaryVerdict{Decision: CanaryPromote, Reasons: notes}
}
// judgeErrors fails when the canary logged more than twice the baseline's
// error lines plus ten. Each count is scaled up to a full day when the log
// error lines plus ten. Each count is scaled up to a full window when the log
// covers only part of its window.
func judgeErrors(baseline, canary CanaryWindowSignals) (string, bool) {
func judgeErrors(baseline, canary CanaryWindowSignals, window time.Duration) (string, bool) {
if baseline.Covered < minLogCoverage || canary.Covered <= 0 {
return "error log doesn't reach back over the baseline; error signal skipped", true
}
scale := func(w CanaryWindowSignals) float64 {
return float64(w.Errors) * float64(CanaryWindow) / float64(w.Covered)
return float64(w.Errors) * float64(window) / float64(w.Covered)
}
before, after := scale(baseline), scale(canary)
limit := 2*before + 10
if after > limit {
return fmt.Sprintf("%.0f error log lines per day against a limit of %.0f (baseline %.0f)", after, limit, before), false
return fmt.Sprintf("%.0f error log lines per window against a limit of %.0f (baseline %.0f)", after, limit, before), false
}
return fmt.Sprintf("%.0f error log lines per day, limit %.0f", after, limit), true
return fmt.Sprintf("%.0f error log lines per window, limit %.0f", after, limit), true
}
// judgeParks fails when the share of finished tasks that parked or died rose
// more than ten points over the baseline.
// more than ten points over the baseline. Both windows need the minimum
// finished tasks before the rates are compared.
func judgeParks(baseline, canary CanaryWindowSignals) (string, bool) {
if canary.Finished < minFinished {
return fmt.Sprintf("%d finished tasks, fewer than %d; park signal skipped", canary.Finished, minFinished), true
if canary.Finished < minFinished || baseline.Finished < minFinished {
return fmt.Sprintf("%d finished tasks on the canary and %d in the baseline, fewer than %d each; park signal skipped", canary.Finished, baseline.Finished, minFinished), true
}
rate := func(w CanaryWindowSignals) float64 {
if w.Finished == 0 {
+29 -26
View File
@@ -32,44 +32,47 @@ var (
const updateSettingsPath = "/api/control-plane/resources/update-settings"
// Signals reads the canary's current state. Tasks and error lines are split
// into the day before the release was issued and the time since.
func (d CanaryDashboard) Signals(ctx context.Context, issued, now time.Time) (CanarySignals, error) {
// into the window before the release was installed on the canary and the time
// since.
func (d CanaryDashboard) Signals(ctx context.Context, installedAt, now time.Time, window time.Duration) (CanarySignals, error) {
var s CanarySignals
var err error
if s.Installed, _, err = d.installed(ctx); err != nil {
if s.Installed, _, _, err = d.Installed(ctx); err != nil {
return s, err
}
if s.Ready, err = d.ready(ctx); err != nil {
return s, err
}
if s.NewActionAlerts, err = d.actionAlerts(ctx, issued); err != nil {
if s.NewActionAlerts, err = d.actionAlerts(ctx, installedAt); err != nil {
return s, err
}
if err := d.tasks(ctx, issued, &s); err != nil {
if err := d.tasks(ctx, installedAt, window, &s); err != nil {
return s, err
}
return s, d.errors(ctx, issued, now, &s)
return s, d.errors(ctx, installedAt, now, window, &s)
}
// installed returns the daemon version the canary runs, and the raw update
// snapshot it offers, which an install request must send back unchanged.
func (d CanaryDashboard) installed(ctx context.Context) (string, json.RawMessage, error) {
// Installed returns the daemon version the canary runs, when that release was
// installed on the canary host (zero when the host doesn't report it), and
// the raw update snapshot it offers, which an install request must send back
// unchanged.
func (d CanaryDashboard) Installed(ctx context.Context) (version string, installedAt time.Time, snapshot json.RawMessage, err error) {
var body struct {
Snapshot json.RawMessage `json:"snapshot"`
}
if _, err := d.call(ctx, http.MethodGet, "/api/chat/update", nil, &body); err != nil {
return "", nil, err
return "", time.Time{}, nil, err
}
var snapshot Snapshot
if err := json.Unmarshal(body.Snapshot, &snapshot); err != nil {
return "", nil, fmt.Errorf("decode update snapshot: %w", err)
var snap Snapshot
if err := json.Unmarshal(body.Snapshot, &snap); err != nil {
return "", time.Time{}, nil, fmt.Errorf("decode update snapshot: %w", err)
}
for _, c := range snapshot.Components {
for _, c := range snap.Components {
if c.ID == "daemon" {
return c.Installed, body.Snapshot, nil
return c.Installed, snap.InstalledAt, body.Snapshot, nil
}
}
return "", body.Snapshot, errors.New("canary reports no daemon version")
return "", time.Time{}, body.Snapshot, errors.New("canary reports no daemon version")
}
func (d CanaryDashboard) ready(ctx context.Context) (bool, error) {
@@ -104,8 +107,8 @@ func (d CanaryDashboard) actionAlerts(ctx context.Context, since time.Time) ([]s
// tasks reads tasks, newest update first, until it reaches tasks older than
// the baseline window.
func (d CanaryDashboard) tasks(ctx context.Context, issued time.Time, s *CanarySignals) error {
from := issued.Add(-CanaryWindow)
func (d CanaryDashboard) tasks(ctx context.Context, at time.Time, window time.Duration, s *CanarySignals) error {
from := at.Add(-window)
cursor := ""
for range 50 {
var page []struct {
@@ -125,7 +128,7 @@ func (d CanaryDashboard) tasks(ctx context.Context, issued time.Time, s *CanaryS
continue
}
w := &s.Canary
if t.UpdatedAt.Before(issued) {
if t.UpdatedAt.Before(at) {
w = &s.Baseline
}
w.Finished++
@@ -144,7 +147,7 @@ func (d CanaryDashboard) tasks(ctx context.Context, issued time.Time, s *CanaryS
// errors counts error-level lines in each window. A log read returns a
// limited number of lines, so a truncated read only reaches back as far as
// its oldest line.
func (d CanaryDashboard) errors(ctx context.Context, issued, now time.Time, s *CanarySignals) error {
func (d CanaryDashboard) errors(ctx context.Context, at, now time.Time, window time.Duration, s *CanarySignals) error {
var body struct {
Entries []struct {
Time time.Time `json:"time"`
@@ -154,24 +157,24 @@ func (d CanaryDashboard) errors(ctx context.Context, issued, now time.Time, s *C
if _, err := d.call(ctx, http.MethodGet, "/api/logs?level=error&limit=2000", nil, &body); err != nil {
return err
}
from := issued.Add(-CanaryWindow)
from := at.Add(-window)
if body.Truncated {
from = now
if len(body.Entries) > 0 {
from = later(body.Entries[0].Time, issued.Add(-CanaryWindow))
from = later(body.Entries[0].Time, at.Add(-window))
}
}
for _, e := range body.Entries {
switch {
case e.Time.Before(from):
case e.Time.Before(issued):
case e.Time.Before(at):
s.Baseline.Errors++
default:
s.Canary.Errors++
}
}
s.Baseline.Covered = max(issued.Sub(from), 0)
s.Canary.Covered = now.Sub(later(from, issued))
s.Baseline.Covered = max(at.Sub(from), 0)
s.Canary.Covered = now.Sub(later(from, at))
return nil
}
@@ -232,7 +235,7 @@ func (d CanaryDashboard) Install(ctx context.Context, version string, wait time.
deadline := time.Now().Add(wait)
requested := false
for {
installed, snapshot, err := d.installed(ctx)
installed, _, snapshot, err := d.Installed(ctx)
switch {
case err != nil:
// The dashboard drops off while the install restarts it.
+51 -18
View File
@@ -5,10 +5,11 @@ import (
"time"
)
// A release promotes only after a full window with no hard failure and no
// relative signal past its threshold; any hard failure after settling rolls
// it back at once.
// A release promotes only after a full window counted from its install with
// no hard failure and no relative signal past its threshold; any failure after
// settling rolls it back at once, even mid-window.
func TestJudgeCanary(t *testing.T) {
installed := time.Date(2026, 1, 2, 12, 0, 0, 0, time.UTC)
healthy := func() CanarySignals {
return CanarySignals{
Installed: "1.1.0", Ready: true,
@@ -23,34 +24,66 @@ func TestJudgeCanary(t *testing.T) {
}
cases := []struct {
name string
elapsed time.Duration
elapsed time.Duration // now minus the install time
window time.Duration
signals CanarySignals
want CanaryDecision
}{
{"settling ignores a failed check", 10 * time.Minute, with(func(s *CanarySignals) { s.Ready = false }), CanaryWait},
{"healthy mid-window waits", 2 * time.Hour, healthy(), CanaryWait},
{"healthy at the window promotes", CanaryWindow, healthy(), CanaryPromote},
{"the watchdog rolled the install back", 2 * time.Hour, with(func(s *CanarySignals) { s.Installed = "1.0.0" }), CanaryRollBack},
{"readiness degraded", 2 * time.Hour, with(func(s *CanarySignals) { s.Ready = false }), CanaryRollBack},
{"a new action alert", 2 * time.Hour, with(func(s *CanarySignals) { s.NewActionAlerts = []string{"queue stalled"} }), CanaryRollBack},
{"errors past twice the baseline plus ten", CanaryWindow, with(func(s *CanarySignals) { s.Canary.Errors = 31 }), CanaryRollBack},
{"errors at the limit", CanaryWindow, with(func(s *CanarySignals) { s.Canary.Errors = 30 }), CanaryPromote},
{"errors scaled from a partly covered log", CanaryWindow, with(func(s *CanarySignals) {
{"settling counts from the install", 10 * time.Minute, CanaryWindow, with(func(s *CanarySignals) { s.Ready = false }), CanaryWait},
{"healthy mid-window waits", 2 * time.Hour, CanaryWindow, healthy(), CanaryWait},
{"healthy at the window promotes", CanaryWindow, CanaryWindow, healthy(), CanaryPromote},
{"healthy at a custom window promotes", 12 * time.Hour, 12 * time.Hour, healthy(), CanaryPromote},
{"the watchdog rolled the install back", 2 * time.Hour, CanaryWindow, with(func(s *CanarySignals) { s.Installed = "1.0.0" }), CanaryRollBack},
{"readiness degraded", 2 * time.Hour, CanaryWindow, with(func(s *CanarySignals) { s.Ready = false }), CanaryRollBack},
{"a new action alert", 2 * time.Hour, CanaryWindow, with(func(s *CanarySignals) { s.NewActionAlerts = []string{"queue stalled"} }), CanaryRollBack},
{"errors past twice the baseline plus ten", CanaryWindow, CanaryWindow, with(func(s *CanarySignals) { s.Canary.Errors = 31 }), CanaryRollBack},
{"errors at the limit", CanaryWindow, CanaryWindow, with(func(s *CanarySignals) { s.Canary.Errors = 30 }), CanaryPromote},
{"errors scaled from a partly covered log", CanaryWindow, CanaryWindow, with(func(s *CanarySignals) {
s.Canary.Errors, s.Canary.Covered = 8, 6*time.Hour
}), CanaryRollBack},
{"a baseline the log doesn't reach skips errors", CanaryWindow, with(func(s *CanarySignals) {
{"a baseline the log doesn't reach skips errors", CanaryWindow, CanaryWindow, with(func(s *CanarySignals) {
s.Baseline.Covered, s.Canary.Errors = 30*time.Minute, 500
}), CanaryPromote},
{"park rate up more than ten points", CanaryWindow, with(func(s *CanarySignals) { s.Canary.Bad = 5 }), CanaryRollBack},
{"park rate up ten points", CanaryWindow, with(func(s *CanarySignals) { s.Canary.Bad = 4 }), CanaryPromote},
{"too few tasks skips the park rate", CanaryWindow, with(func(s *CanarySignals) { s.Canary = CanaryWindowSignals{Finished: 4, Bad: 4, Covered: CanaryWindow} }), CanaryPromote},
{"a mid-window error breach rolls back early", 2 * time.Hour, CanaryWindow, with(func(s *CanarySignals) { s.Canary.Errors = 31 }), CanaryRollBack},
{"park rate up more than ten points", CanaryWindow, CanaryWindow, with(func(s *CanarySignals) { s.Canary.Bad = 5 }), CanaryRollBack},
{"park rate up ten points", CanaryWindow, CanaryWindow, with(func(s *CanarySignals) { s.Canary.Bad = 4 }), CanaryPromote},
{"too few canary tasks skips the park rate", CanaryWindow, CanaryWindow, with(func(s *CanarySignals) { s.Canary = CanaryWindowSignals{Finished: 4, Bad: 4, Covered: CanaryWindow} }), CanaryPromote},
{"a thin baseline skips the park rate", CanaryWindow, CanaryWindow, with(func(s *CanarySignals) {
s.Baseline.Finished, s.Baseline.Bad = 2, 0
s.Canary.Bad = 5
}), CanaryPromote},
{"a mid-window park breach rolls back early", 2 * time.Hour, CanaryWindow, with(func(s *CanarySignals) { s.Canary.Bad = 5 }), CanaryRollBack},
}
for _, tc := range cases {
t.Run(tc.name, func(t *testing.T) {
got := JudgeCanary("1.1.0", tc.elapsed, tc.signals)
got := JudgeCanary("1.1.0", installed, installed.Add(tc.elapsed), tc.window, tc.signals)
if got.Decision != tc.want {
t.Fatalf("decision = %s, want %s (%v)", got.Decision, tc.want, got.Reasons)
}
})
}
}
// Judging "now" skips the settle and window waits but still applies every
// hard and relative check.
func TestJudgeCanaryNow(t *testing.T) {
healthy := CanarySignals{
Installed: "1.1.0", Ready: true,
Baseline: CanaryWindowSignals{Finished: 20, Bad: 2, Errors: 10, Covered: CanaryWindow},
Canary: CanaryWindowSignals{Finished: 20, Bad: 2, Errors: 12, Covered: CanaryWindow},
}
now := time.Date(2026, 1, 2, 12, 10, 0, 0, time.UTC)
if got := JudgeCanaryNow("1.1.0", now, CanaryWindow, healthy); got.Decision != CanaryPromote {
t.Fatalf("healthy now = %s, want promote (%v)", got.Decision, got.Reasons)
}
degraded := healthy
degraded.Ready = false
if got := JudgeCanaryNow("1.1.0", now, CanaryWindow, degraded); got.Decision != CanaryRollBack {
t.Fatalf("degraded now = %s, want roll back (%v)", got.Decision, got.Reasons)
}
breached := healthy
breached.Canary.Bad = 10
if got := JudgeCanaryNow("1.1.0", now, CanaryWindow, breached); got.Decision != CanaryRollBack {
t.Fatalf("park breach now = %s, want roll back (%v)", got.Decision, got.Reasons)
}
}
+47
View File
@@ -0,0 +1,47 @@
package releaseupdate
import (
"context"
"fmt"
digest "github.com/opencontainers/go-digest"
ocispec "github.com/opencontainers/image-spec/specs-go/v1"
"oras.land/oras-go/v2/registry"
"github.com/samcharles93/archie-core/internal/domain/distribution"
)
// Installable reports whether a release can still be installed from source:
// the zip blob its statement names must be present in the release repository
// and the agent image it pins must still resolve. A canary must never be
// pinned to a release whose artifacts are gone, or it would be stranded on a
// version it can never install again.
func Installable(ctx context.Context, source distribution.Source, release Release) error {
zipDigest, err := digest.Parse(release.Statement.Zip.Digest)
if err != nil {
return fmt.Errorf("zip digest %q: %w", release.Statement.Zip.Digest, err)
}
repo, err := pushRepository(source.Reference)
if err != nil {
return err
}
exists, err := repo.Exists(ctx, ocispec.Descriptor{Digest: zipDigest})
if err != nil {
return fmt.Errorf("resolve zip blob %s in %s: %w", release.Statement.Zip.Digest, source.Reference, err)
}
if !exists {
return fmt.Errorf("zip blob %s (%s) is absent from %s", release.Statement.Zip.Digest, release.Statement.Zip.Name, source.Reference)
}
image, err := registry.ParseReference(release.Statement.AgentImage)
if err != nil {
return fmt.Errorf("agent image %q: %w", release.Statement.AgentImage, err)
}
imageRepo, err := pushRepository(image.Registry + "/" + image.Repository)
if err != nil {
return err
}
if _, err := imageRepo.Resolve(ctx, image.Reference); err != nil {
return fmt.Errorf("agent image %s: %w", release.Statement.AgentImage, err)
}
return nil
}
+3
View File
@@ -44,6 +44,9 @@ type Component struct {
type Snapshot struct {
Components []Component
Deferred bool
// InstalledAt is when the installed release was committed on this host,
// from its installed.json. Zero when the host doesn't record one.
InstalledAt time.Time `json:"installed_at"`
}
func (s Snapshot) Available() []Component {
+54
View File
@@ -40,6 +40,7 @@ const (
journalDir = "transaction"
journalFile = "journal.json"
snapshotFile = "db.pre-update.dump"
installedFile = "installed.json"
lockWait = 60 * time.Second
)
@@ -126,6 +127,39 @@ type journal struct {
// JournalPath is where an unfinished transaction lives.
func JournalPath(stateDir string) string { return filepath.Join(stateDir, journalDir) }
// installedRecord is the host's durable note of which release it last
// committed (or rolled back to) and when, so a canary can count its window
// from the actual install rather than the release statement.
type installedRecord struct {
Version string `json:"version"`
CommittedAt time.Time `json:"committed_at"`
}
// recordInstall writes installed.json atomically: the file is the source of
// truth for when the running release was installed.
func recordInstall(stateDir, version string, at time.Time) error {
data, err := json.Marshal(installedRecord{Version: version, CommittedAt: at.UTC()})
if err != nil {
return err
}
return fsutil.WriteFileAtomic(filepath.Join(stateDir, installedFile), data, 0o600)
}
// LastInstall reads the host's installed record. ok is false when the file is
// missing or malformed, as on a host that updated before install times were
// recorded.
func LastInstall(stateDir string) (version string, at time.Time, ok bool) {
data, err := os.ReadFile(filepath.Join(stateDir, installedFile))
if err != nil {
return "", time.Time{}, false
}
var record installedRecord
if err := json.Unmarshal(data, &record); err != nil || record.Version == "" || record.CommittedAt.IsZero() {
return "", time.Time{}, false
}
return record.Version, record.CommittedAt, true
}
// Run drives the transaction in the journal to an outcome. recovering is set
// on boot: it never waits for the lock, and a transaction that was prepared
// but never started is abandoned rather than started unattended.
@@ -158,6 +192,17 @@ func (t *Transaction) Run(ctx context.Context, recovering bool) error {
if err := t.startUnits(ctx, j, ""); err != nil {
return err
}
// A crash between the outcome phase and recordInstall must not lose
// the install record, so the recovery path writes it too.
version := j.Version
if j.Phase == phaseRolledBack {
version = j.Previous[ComponentDaemon]
}
if version != "" {
if err := recordInstall(t.StateDir, version, time.Now().UTC()); err != nil {
return err
}
}
return t.remove()
default:
return t.rollback(ctx, j)
@@ -182,6 +227,9 @@ func (t *Transaction) promote(ctx context.Context, j journal) error {
if err := t.setPhase(&j, phaseCommitted); err != nil {
return err
}
if err := recordInstall(t.StateDir, j.Version, time.Now().UTC()); err != nil {
return fmt.Errorf("record the install: %w", err)
}
return t.remove()
}
@@ -252,6 +300,12 @@ func (t *Transaction) rollback(ctx context.Context, j journal) error {
if err := t.setPhase(&j, phaseRolledBack); err != nil {
return t.incomplete(err)
}
// The host is back on the previous release; installed.json follows it.
if version := j.Previous[ComponentDaemon]; version != "" {
if err := recordInstall(t.StateDir, version, time.Now().UTC()); err != nil {
return t.incomplete(err)
}
}
if err := t.startUnits(ctx, j, started); err != nil {
return t.incomplete(err)
}
@@ -21,6 +21,7 @@ import (
digest "github.com/opencontainers/go-digest"
ocispec "github.com/opencontainers/image-spec/specs-go/v1"
"github.com/samcharles93/archie-core/internal/buildinfo"
"github.com/samcharles93/archie-core/internal/domain/distribution"
"github.com/samcharles93/archie-core/internal/infrastructure/signedoci"
)
@@ -146,6 +147,11 @@ func TestTransaction(t *testing.T) {
// the env file beside its config; pin both to this test's own files so
// the result never depends on the machine's real archie config.
t.Setenv(DatabasePasswordEnv, "test-password")
// Stamp this binary as a release, as a host installed from one is, so the
// journal's previous daemon version is a real version.
stamped := buildinfo.Version
buildinfo.Version = "1.0.0"
defer func() { buildinfo.Version = stamped }()
configPath := filepath.Join(t.TempDir(), "config.toml")
if err := os.WriteFile(configPath, []byte("database_url = \"postgres://archie@/archie?host=/tmp/archie-test\"\n"), 0o600); err != nil {
t.Fatal(err)
@@ -257,10 +263,51 @@ func TestTransaction(t *testing.T) {
found && (got.HealthCheck != tt.want.report || got.RolledBack != (tt.want.report == "failed") || got.RestoredDatabase != tt.want.restoredDatabase) {
t.Errorf("report = %+v (found %v), want %+v", got, found, tt.want)
}
// A transaction that ran to an outcome records which release the
// host now runs and when it was installed.
if tt.prepare == "" && tt.crashAt == "" {
if version, at, ok := LastInstall(tx.StateDir); !ok || version != tt.want.version || at.IsZero() {
t.Errorf("installed.json after the transaction = %q, %v, %v; want %s", version, at, ok, tt.want.version)
}
}
})
}
}
// installed.json is the host's durable record of which release it last
// committed or rolled back to, and when.
func TestLastInstall(t *testing.T) {
dir := t.TempDir()
if _, _, ok := LastInstall(dir); ok {
t.Fatal("a missing installed.json reports ok")
}
at := time.Date(2026, 1, 2, 3, 4, 5, 0, time.UTC)
if err := recordInstall(dir, "1.2.3", at); err != nil {
t.Fatal(err)
}
var raw map[string]any
data, err := os.ReadFile(filepath.Join(dir, installedFile))
if err != nil {
t.Fatal(err)
}
if err := json.Unmarshal(data, &raw); err != nil {
t.Fatalf("installed.json is not JSON: %v", err)
}
if raw["version"] != "1.2.3" || raw["committed_at"] != "2026-01-02T03:04:05Z" {
t.Fatalf("installed.json = %v, want version 1.2.3 and committed_at 2026-01-02T03:04:05Z", raw)
}
version, got, ok := LastInstall(dir)
if !ok || version != "1.2.3" || !got.Equal(at) {
t.Fatalf("LastInstall = %q, %v, %v; want 1.2.3, %v, true", version, got, ok, at)
}
if err := os.WriteFile(filepath.Join(dir, installedFile), []byte("{not json"), 0o600); err != nil {
t.Fatal(err)
}
if _, _, ok := LastInstall(dir); ok {
t.Fatal("a malformed installed.json reports ok")
}
}
// releaseHandler serves one release's zip and SHA256SUMS the way GitHub
// release downloads are laid out.
func releaseHandler(t *testing.T, candidate string, corrupt bool) (http.Handler, ed25519.PublicKey) {
@@ -13,8 +13,8 @@ const defaultDenialsListLimit = 100
// first. The route's read_audit grant limits it to the org's admins and
// owners, and the store scopes the list by the credential's org, so no
// request field chooses the org an operator reads.
func (s *Server) handleListDenials(w http.ResponseWriter, r *http.Request) {
if s.DenialLister == nil {
func (a *accessAPI) handleListDenials(w http.ResponseWriter, r *http.Request) {
if a.denials == nil {
http.Error(w, "denial records unavailable", http.StatusServiceUnavailable)
return
}
@@ -22,7 +22,7 @@ func (s *Server) handleListDenials(w http.ResponseWriter, r *http.Request) {
if err != nil || limit <= 0 || limit > 1000 {
limit = defaultDenialsListLimit
}
denials, err := s.DenialLister.ListDenials(r.Context(), policyOrg(r), int(limit))
denials, err := a.denials.ListDenials(r.Context(), policyOrg(r), int(limit))
if err != nil {
writeControlPlaneError(w, err)
return
@@ -95,7 +95,7 @@ func TestHandleListDenialsScopesToCallersOrg(t *testing.T) {
ctx := org.WithOrg(access.WithPrincipal(t.Context(), principal), tt.caller)
req := httptest.NewRequestWithContext(ctx, http.MethodGet, "/api/access/denials"+tt.query, nil)
rec := httptest.NewRecorder()
srv.handleListDenials(rec, req)
newAccessAPI(srv).handleListDenials(rec, req)
if rec.Code != http.StatusOK {
t.Fatalf("status = %d, want 200 body=%s", rec.Code, rec.Body.String())
}
@@ -129,7 +129,7 @@ func TestHandleListDenialsUnwired(t *testing.T) {
srv := &Server{}
req := httptest.NewRequestWithContext(t.Context(), http.MethodGet, "/api/access/denials", nil)
rec := httptest.NewRecorder()
srv.handleListDenials(rec, req)
newAccessAPI(srv).handleListDenials(rec, req)
if rec.Code != http.StatusServiceUnavailable {
t.Fatalf("status = %d, want 503", rec.Code)
}
@@ -9,6 +9,24 @@ import (
"github.com/samcharles93/archie-core/internal/domain/org"
)
// accessAPI owns the /api/access routes: the stored policy chain and the
// recorded denials. Either store may be nil; its routes then answer 503.
type accessAPI struct {
policies access.PolicyStore
denials access.DenialStore
}
func newAccessAPI(s *Server) *accessAPI {
return &accessAPI{policies: s.Policies, denials: s.DenialLister}
}
func (a *accessAPI) register(mux *http.ServeMux) {
mux.HandleFunc("GET /api/access/policies", a.handleListPolicies)
mux.HandleFunc("PUT /api/access/policies", a.handlePutPolicy)
mux.HandleFunc("DELETE /api/access/policies", a.handleDeletePolicy)
mux.HandleFunc("GET /api/access/denials", a.handleListDenials)
}
// policyOrg is the org a policy request acts in: the caller's.
func policyOrg(r *http.Request) org.OrgID {
if principal, ok := access.PrincipalFromContext(r.Context()); ok {
@@ -19,12 +37,12 @@ func policyOrg(r *http.Request) org.OrgID {
// handleListPolicies lists the caller's org policies, and the instance
// policies when the caller is the instance owner.
func (s *Server) handleListPolicies(w http.ResponseWriter, r *http.Request) {
if s.Policies == nil {
func (a *accessAPI) handleListPolicies(w http.ResponseWriter, r *http.Request) {
if a.policies == nil {
http.Error(w, "policies unavailable", http.StatusServiceUnavailable)
return
}
policies, err := s.Policies.ListPolicies(r.Context())
policies, err := a.policies.ListPolicies(r.Context())
if err != nil {
writeControlPlaneError(w, err)
return
@@ -39,12 +57,12 @@ func (s *Server) handleListPolicies(w http.ResponseWriter, r *http.Request) {
writeJSON(w, map[string]any{"policies": out, "shipped": shippedPolicyIDs()})
}
func (s *Server) handlePutPolicy(w http.ResponseWriter, r *http.Request) {
p, ok := s.decodePolicy(w, r)
func (a *accessAPI) handlePutPolicy(w http.ResponseWriter, r *http.Request) {
p, ok := a.decodePolicy(w, r)
if !ok {
return
}
version, err := s.Policies.PutPolicy(r.Context(), p)
version, err := a.policies.PutPolicy(r.Context(), p)
if err != nil {
writePolicyError(w, err)
return
@@ -52,12 +70,12 @@ func (s *Server) handlePutPolicy(w http.ResponseWriter, r *http.Request) {
writeJSON(w, map[string]int64{"version": version})
}
func (s *Server) handleDeletePolicy(w http.ResponseWriter, r *http.Request) {
p, ok := s.decodePolicy(w, r)
func (a *accessAPI) handleDeletePolicy(w http.ResponseWriter, r *http.Request) {
p, ok := a.decodePolicy(w, r)
if !ok {
return
}
if err := s.Policies.DeletePolicy(r.Context(), p); err != nil {
if err := a.policies.DeletePolicy(r.Context(), p); err != nil {
writePolicyError(w, err)
return
}
@@ -66,8 +84,8 @@ func (s *Server) handleDeletePolicy(w http.ResponseWriter, r *http.Request) {
// decodePolicy reads a policy and pins it to the caller's org: no request
// field chooses the org, and instance policies are not edited here.
func (s *Server) decodePolicy(w http.ResponseWriter, r *http.Request) (access.Policy, bool) {
if s.Policies == nil {
func (a *accessAPI) decodePolicy(w http.ResponseWriter, r *http.Request) (access.Policy, bool) {
if a.policies == nil {
http.Error(w, "policies unavailable", http.StatusServiceUnavailable)
return access.Policy{}, false
}
+18 -1
View File
@@ -58,8 +58,12 @@ func (s *Server) accountIdentity(r *http.Request) (identity.Identity, error) {
func (s *Server) requireAccount(h http.Handler) http.Handler {
return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
// Any active identity may authenticate: a person, or a service
// account whose bearer token carries a scoped grant. What the
// identity may then do is the policy chain's call, not the
// credential kind's; an inactive identity is refused here.
value, err := s.accountIdentity(r)
if err != nil || value.Kind != identity.KindUser || !value.CanAct() {
if err != nil || !value.CanAct() {
if err == nil {
err = identity.ErrCredentialRejected
}
@@ -413,3 +417,16 @@ func (s *Server) authenticateBrowser(ctx context.Context, credential string) (id
}
return verify(ctx, credential)
}
func (s *Server) memberIssuer(r *http.Request) string {
if store, ok := s.authenticationSettings(); ok {
value, err := store.AuthenticationSettings(r.Context())
if err != nil {
return ""
}
if value.PublicURL != "" {
return value.OIDCIssuer
}
}
return s.Issuer
}
+144
View File
@@ -2,6 +2,7 @@ package webui
import (
"context"
"encoding/json"
"errors"
"net/http"
"net/http/httptest"
@@ -131,6 +132,149 @@ func TestAccountSessionFailureIsClosed(t *testing.T) {
}
}
// TestRequireAccountAdmitsAnyActiveIdentity pins the bearer-token gate: any
// active identity -- a person or a service account -- is admitted and rides
// the context as the acting identity, while an inactive one is refused, so a
// scoped service-account token can use the API without being a user.
func TestRequireAccountAdmitsAnyActiveIdentity(t *testing.T) {
tests := []struct {
name string
value identity.Identity
want int
}{
{"active user", identity.Identity{ID: "u1", Kind: identity.KindUser, DisplayName: "User", Lifecycle: identity.LifecycleActive}, http.StatusOK},
{"active service account", identity.Identity{ID: "sa1", Kind: identity.KindServiceAccount, DisplayName: "Canary", Lifecycle: identity.LifecycleActive}, http.StatusOK},
{"active bot", identity.Identity{ID: "b1", Kind: identity.KindBot, DisplayName: "Bot", Lifecycle: identity.LifecycleActive}, http.StatusOK},
{"suspended user", identity.Identity{ID: "u2", Kind: identity.KindUser, DisplayName: "User", Lifecycle: identity.LifecycleSuspended}, http.StatusUnauthorized},
{"retired service account", identity.Identity{ID: "sa2", Kind: identity.KindServiceAccount, DisplayName: "Canary", Lifecycle: identity.LifecycleRetired}, http.StatusUnauthorized},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
srv := &Server{
Authenticate: func(context.Context, string) (identity.Identity, error) { return tt.value, nil },
}
var acting identity.IdentityID
handler := srv.requireAccount(http.HandlerFunc(func(_ http.ResponseWriter, r *http.Request) {
if value, ok := ActingIdentity(r.Context()); ok {
acting = value.ID
}
}))
req := httptest.NewRequestWithContext(t.Context(), http.MethodGet, "https://archie.example/api/summary", nil)
req.Header.Set("Authorization", "Bearer archie_pat_test")
rec := httptest.NewRecorder()
handler.ServeHTTP(rec, req)
if rec.Code != tt.want {
t.Fatalf("status=%d want=%d body=%s", rec.Code, tt.want, rec.Body.String())
}
if tt.want == http.StatusOK && acting != tt.value.ID {
t.Fatalf("acting identity = %q, want %q", acting, tt.value.ID)
}
})
}
}
// identityRepo answers Get for the identities a test names.
type identityRepo struct {
identity.Repository
values map[identity.IdentityID]identity.Identity
}
func (s identityRepo) Get(_ context.Context, id identity.IdentityID) (identity.Identity, error) {
value, ok := s.values[id]
if !ok {
return identity.Identity{}, identity.ErrNotFound
}
return value, nil
}
// identityTokenStore carries both token contracts: the remote caller's view
// (embedded) and the store side, whose BindPersonalToken names an identity.
type identityTokenStore struct {
identity.PersonalTokens
bound bool
id identity.IdentityID
subject identity.Subject
name string
}
func (s *identityTokenStore) BindPersonalToken(_ context.Context, id identity.IdentityID, subject identity.Subject, name string, _ *time.Time, _ identity.Audit) error {
s.bound, s.id, s.subject, s.name = true, id, subject, name
return nil
}
func (s *identityTokenStore) SubjectsOf(context.Context, identity.IdentityID, string) ([]identity.PersonalToken, error) {
return nil, nil
}
func (s *identityTokenStore) UnbindSubject(context.Context, identity.IdentityID, identity.Subject, identity.Audit) error {
return nil
}
func (s *identityTokenStore) PersonalTokenOf(context.Context, identity.Subject) (identity.PersonalToken, error) {
return identity.PersonalToken{}, identity.ErrNotFound
}
func (s *identityTokenStore) TouchPersonalToken(context.Context, identity.Subject, time.Time) error {
return nil
}
// TestIdentityTokenCreate pins the per-identity token route: an active
// service account gets its raw token exactly once and the store binds only
// its hash, while an unknown or inactive identity is refused.
func TestIdentityTokenCreate(t *testing.T) {
active := identity.Identity{ID: "sa1", Kind: identity.KindServiceAccount, DisplayName: "Canary", Lifecycle: identity.LifecycleActive}
suspended := identity.Identity{ID: "sa2", Kind: identity.KindServiceAccount, DisplayName: "Old", Lifecycle: identity.LifecycleSuspended}
tests := []struct {
name, id, body string
want int
}{
{"active service account", "sa1", `{"name":"ci","days":30}`, http.StatusCreated},
{"unknown identity", "nope", `{"name":"ci","days":30}`, http.StatusBadRequest},
{"inactive identity", "sa2", `{"name":"ci","days":30}`, http.StatusConflict},
{"bad name", "sa1", `{"name":" ","days":30}`, http.StatusBadRequest},
{"bad lifetime", "sa1", `{"name":"ci","days":10}`, http.StatusBadRequest},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
store := &identityTokenStore{}
srv := &Server{
Identities: identityRepo{values: map[identity.IdentityID]identity.Identity{active.ID: active, suspended.ID: suspended}},
PersonalTokens: store,
}
req := httptest.NewRequestWithContext(t.Context(), http.MethodPost, "https://archie.example/api/identities/"+tt.id+"/tokens", strings.NewReader(tt.body))
req.Header.Set("Content-Type", "application/json")
req.SetPathValue("id", tt.id)
rec := httptest.NewRecorder()
newIdentitiesAPI(srv).handleIdentityTokenCreate(rec, req)
if rec.Code != tt.want {
t.Fatalf("status=%d want=%d body=%s", rec.Code, tt.want, rec.Body.String())
}
if tt.want != http.StatusCreated {
if store.bound {
t.Fatal("a refused request still bound a token")
}
return
}
var body struct {
ID string `json:"id"`
Token string `json:"token"`
}
if err := json.Unmarshal(rec.Body.Bytes(), &body); err != nil {
t.Fatalf("response: %v (%s)", err, rec.Body.String())
}
if !strings.HasPrefix(body.Token, "archie_pat_") || body.ID == "" {
t.Fatalf("response must show the raw token once: %+v", body)
}
if !store.bound || store.id != active.ID || store.name != "ci" {
t.Fatalf("bound = %+v, want id %q name %q", store, active.ID, "ci")
}
if store.subject.Issuer != identity.PersonalTokenIssuer || store.subject.Subject != body.ID {
t.Fatalf("store kept subject %+v, want the token's hash %q under %q", store.subject, body.ID, identity.PersonalTokenIssuer)
}
})
}
}
type unavailableAccounts struct{ identity.Accounts }
func (unavailableAccounts) AccountSetupRequired(context.Context) (bool, error) {
@@ -3,6 +3,7 @@ package webui
import (
"context"
"errors"
"log/slog"
"net/http"
"strconv"
"strings"
@@ -72,10 +73,23 @@ func viewOfAlert(a alert.Alert) alertView {
return view
}
// alertsAPI owns the /api/alerts routes. A nil source answers 503.
type alertsAPI struct {
alerts AlertSource
log *slog.Logger
}
func newAlertsAPI(s *Server) *alertsAPI { return &alertsAPI{alerts: s.Alerts, log: s.Log} }
func (a *alertsAPI) register(mux *http.ServeMux) {
mux.HandleFunc("GET /api/alerts", a.handleAlerts)
mux.HandleFunc("POST /api/alerts/{id}/dismiss", a.handleAlertDismiss)
}
// handleAlerts lists the caller's org's alerts: open ones unless ?state names
// others (comma separated), most severe first.
func (s *Server) handleAlerts(w http.ResponseWriter, r *http.Request) {
if s.Alerts == nil {
func (a *alertsAPI) handleAlerts(w http.ResponseWriter, r *http.Request) {
if a.alerts == nil {
http.Error(w, "alerts unavailable", http.StatusServiceUnavailable)
return
}
@@ -85,9 +99,9 @@ func (s *Server) handleAlerts(w http.ResponseWriter, r *http.Request) {
filter.States = append(filter.States, alert.State(strings.TrimSpace(state)))
}
}
alerts, err := s.Alerts.ListAlerts(r.Context(), policyOrg(r), filter)
alerts, err := a.alerts.ListAlerts(r.Context(), policyOrg(r), filter)
if err != nil {
s.Log.ErrorContext(r.Context(), "list alerts", "err", err)
a.log.ErrorContext(r.Context(), "list alerts", "err", err)
http.Error(w, "list alerts failed", http.StatusBadGateway)
return
}
@@ -99,8 +113,8 @@ func (s *Server) handleAlerts(w http.ResponseWriter, r *http.Request) {
}
// handleAlertDismiss dismisses one open alert of the caller's org.
func (s *Server) handleAlertDismiss(w http.ResponseWriter, r *http.Request) {
if s.Alerts == nil {
func (a *alertsAPI) handleAlertDismiss(w http.ResponseWriter, r *http.Request) {
if a.alerts == nil {
http.Error(w, "alerts unavailable", http.StatusServiceUnavailable)
return
}
@@ -109,12 +123,12 @@ func (s *Server) handleAlertDismiss(w http.ResponseWriter, r *http.Request) {
http.Error(w, "invalid alert id", http.StatusBadRequest)
return
}
err = s.Alerts.DismissAlert(r.Context(), policyOrg(r), id, access.ActorFromContext(r.Context()))
err = a.alerts.DismissAlert(r.Context(), policyOrg(r), id, access.ActorFromContext(r.Context()))
switch {
case errors.Is(err, alert.ErrNotFound):
http.Error(w, "alert not found", http.StatusNotFound)
case err != nil:
s.Log.ErrorContext(r.Context(), "dismiss alert", "err", err)
a.log.ErrorContext(r.Context(), "dismiss alert", "err", err)
http.Error(w, "dismiss alert failed", http.StatusBadGateway)
default:
w.WriteHeader(http.StatusNoContent)
Loaded 100 of 418 files, more files were not shown because too many files have changed in this diff. Show more