Task agent runs and MCP sampling record against their task (billed to the
task's org); gateway chat turns, titles, sampling and curators record
against the system org. Rows carry source, workflow, step, alias,
provider/model and tokens for later chargeback.
Agent steps, implement, pr-review, MCP sampling, chat, embedding and
transcription resolve a model alias (empty means the purpose's default)
and fail on an unknown alias instead of falling back. The step editor and
chat pick aliases; workflow definitions naming an unknown alias are
refused on save.
internal/domain/workflow imported internal/agentexec for its protocol types,
pulling the AI runtime into every binary that touches workflow types. The
request, result and runner contract now live in internal/domain/agentrun;
agentexec keeps the runtime. The model catalog cache moves from servicekit
into modelcatalog and WaitForPeer into staterpc, so archie-state-store and
archie-messaging no longer link the AI runtime.
Remove PRD/doc paths, bead IDs, phase notes and deleted-test references
from comments; drop test-only hooks left by the test deletion. Includes
in-progress archied bootstrap, gateway and query changes.
The adversarial self-review system was removed in 7299b923, but four pages
describing it survived: the architecture page, its design-decision PRD, and the
two superseded calibration/operator PRDs. The architecture page carried a
hand-written "removed, not shipped" banner and was kept only because live PRDs
linked to it, so every reader paid for a system the tree no longer has.
Delete all four and repoint the live referrers at the authority that replaced
them (docs/prds/pr-review-agent.md): inline-review.md's Builds-on line,
pr-review-remediation.md's isolation-contract citation, pr-review-agent.md's
supersede clause, CLAUDE.md's contributor rule, and three Go comments that cited
the deleted paths. Regenerate docs/data/generated/dev-docs.json.
History stays in git; the tree stops describing a system it does not have.
The baseline stage treated every red gate as a pre-existing code defect and
dispatched a TDD repair agent. A failure the environment caused -- no Docker
daemon for the test harness, or a gate command that is not installed -- cannot
be fixed from the worktree, so the run spent a full agent budget and then
parked. A provider error during that repair run (observed: HTTP 402) replaced
the gate cause in the park reason entirely.
Classify at the producer: gateEnvironmentCause names a failure whose command
could not start (*exec.Error) or whose output names unavailable host
infrastructure (the Docker shapes from archie-core-71mq). That failure parks
with the cause and the gate output and starts no builder. A real code failure
still dispatches exactly one repair run, and every post-dispatch failure now
carries the gate output alongside the agent/provider error instead of masking
it.
Refs archie-core-rwy6.
StagePRReviewAndOpenPR runs the pr-review pipeline's decision phases
(1-8) against the task's own uncommitted change, using a new
localPRSource that reads Diff/Snapshot straight from Trees instead of
fetching an external PR -- there is no PR number yet to fetch one by,
which is why this trigger needed its own PRSource rather than the
forge-backed one afbk.7 already built for external PRs.
An unchallenged blocking finding parks the task with the findings
instead of opening a PR; a pipeline stage failure parks it the same
way any stage failure does. Once the PR opens, advisory findings post
against it -- best-effort, matching the deleted StagePostReviewComments'
convention: a posting failure is logged, not reverted into the outcome
the PR's existence already earned.
Implement() now uses this in place of the plain StageOpenPR. Bootstrap
and TDD are untouched -- the PRD names "the implement workflow"
specifically; extending to the others is a separate call.
Resolves this bead's first blocker (confirmed via read: CompilePinnedWorkflow
returns the same workflow.Workflow type and runs through the same
workflow.Run either compilation path, so a stage setting Outcome ends
the run identically regardless -- stagePROutput's posting logic had to
move inside StageOpenPR's own Run body, not follow it as a separate
Stage, which OpenPR's own doc comment already anticipated).
Still open on this bead: watched-repositories and operator triggers,
both blocked on the credential path for PRSource's real forge-backed
implementation reaching the sandboxed archie-agent process -- see the
bead comment for the concrete design (egress credential-capability
grant, narrow forge diff/archive HTTP methods) not yet built.
docs/prds/pr-review-agent.md's own header already declared this
superseded (docs/architecture/adversarial-review.md, docs/prds/
operator-pr-review.md, docs/prds/review-calibration.md); this executes
that removal per the maintainer's explicit decision, so no reference
to the old path remains:
- internal/domain/workflow/review.go, review_findings.go,
review_report_md.go and their tests: Reviewer, ReviewRequest,
StageReview, StagePostReviewComments, ReviewReport/ReviewFinding and
friends. ReviewComment relocated into workflow.go (still used by
Forger.CreateReviewComments and the new pr-review pipeline).
- internal/app/agentworker/review.go: the ai-sdk-backed subagentReviewer.
- internal/app/archied/pr_review.go: the synchronous operator-review
path (prReviewer, gateway.ChatPRReviewer wiring). The review_pr chat
tool now self-omits (nil reviewer) until the operator trigger is
rebuilt to enqueue an async pr-review task instead of reviewing
synchronously in the chat turn.
- config.Repo.ReviewEnabled and its dashboard exposure.
- Two composition-order tests and one dashboard test that existed only
to pin the deleted path.
Also splits PRReview()'s Stage list into prReviewDecisionStages()
(phases 1-8) and stagePROutput() (phase 9), so a future caller can
splice the decision phases before a PR exists and defer posting until
after -- needed for the "archie's own PRs" trigger, not wired here: it
needs the same stage-loop-around-StageOpenPR semantics this pass
couldn't fully verify in the time available (does an Outcome-setting
stage like StageOpenPR end the compiled-workflow-definition engine the
same way it ends workflow.Run's literal loop?). Carries ScoredFinding.
Adversary through synthesis so that trigger's blocking-finding check
doesn't need to cross-reference back into the pre-synthesis slice.
Not built in this pass, left as a precise continuation (see the bead
comment): the watched-repositories and operator triggers, both blocked
on the same real question -- how a forge-reading credential reaches
PRSource's real implementation when the pr-review pipeline's Stage.Run
bodies execute inside the sandboxed archie-agent process, which holds
none today.
Plan item 4 of docs/prds/execution-tree-state-machine.md (archie-core-1216),
plus the callee half of CancelExecution that plan item 5 deferred until the
tree could decide it.
Every agent call a stage makes records itself: AgentStage's bridge, the
baseline-fix auto-fix run, the adversarial reviewer and a workflow.call each
start a child StepExecution of kind agent or call under the stage's own step
-- TaskContext.StepID, which the engine sets before the stage body starts and
clears when it returns -- and finish it with the call's own outcome and the
tokens it accounted. A child's recording write failure parks the execution
the same way a stage's does: an agent call that ran unrecorded is the
invisibility the step table exists to end. A stage invoked outside the
engine's recorded run -- the stage unit tests -- records no child and runs
unchanged.
The call step is what makes the callee half of the PRD's Cancellation section
decidable, and this lands it: a call step closes when its callee ends, so an
open call step is a callee the caller waits on (wait:true) and a closed one
is a wait:false callee that runs on. CancelExecution reads the open call
steps before its sweep closes them and cancels each named callee through its
own CancelExecution in the caller's write's aftermath -- the callee's own
steps, its own events, its own move to the same destination, and the callee's
own wait:true callees beyond it, the depth the enqueue enforces bounding the
recursion. A callee that cannot be cancelled is reported and the caller's own
record stands.
StartStep refuses a call step whose callee is not this run's own (the
callee's call_parent_task_id must name the recording execution): a
compromised container could otherwise record a step naming a foreign
execution and cancel it. StartStepRequest gains called_execution_id (field 6;
non-breaking, the column and its named consumer already exist) and
StepStart carries it.
PRD decisions recorded here:
- The reviewer records as one agent child named "review": the Reviewer
contract runs one subagent per review, the PRD model's review:1/review:2
are illustrative of depth-1 children. A completed report closes it
succeeded, a not-run review failed with the skip reason; the contract
carries no usage, so tokens are 0.
- A failed-callee call step closes as failed with the callee's detail when
the wait ends otherwise; the single open case is the caller itself being
cancelled, whose CancelExecution already swept this step -- a late write
must not pretend the wait is still the caller's business.
- A call step's callee is cancelled to the same destination the caller's
action named (parked stays parked, closed_wont_do ends the whole chain),
and a callee that cannot be cancelled is reported while the caller's own
record stands either way.
- The agent child's detail is the result summary on success and the error's
text on failure, with the call's own TokensUsed -- the PRD's only usage
field, untracked per call for the reviewer contract.
* feat(workflow): post line-anchored review comments with suggested fixes
Bead archie-core-q9au (docs/prds/inline-review.md). A review's findings reached
the operator only as a flat list in the PR body, so the author had to map every
finding back to its line by hand and apply mechanical fixes by hand. Now the
line-anchored findings are also posted as inline review comments, carrying a
one-click suggestion when the reviewer could state the fix.
Domain (internal/domain/workflow):
- ReviewFinding gains Suggestion; Validate rejects one on anything but a
confirmed finding, so the agent's record_finding call is refused with
something it can act on. Validate splits into validateAnchor /
validateClassification / validateDisposition, which also brings it back under
the cyclop limit the gate enforces.
- StagePostReviewComments runs after StageOpenPR in the implement workflow. It
posts only line-anchored findings (Line > 0); a whole-file finding has no line
to attach to and stays in the body list, which remains the complete record.
It is best-effort: the PR is open and the body already lists every finding, so
a failed comment is logged, never a park -- the same reasoning as OpenPR's
existing best-effort LinkBranch call.
- The suggestion fence is gated on Verdict == confirmed where the fence is
written, because it is a one-click apply: a wrong suggestion breaks the code
and the author has to notice.
Forge (internal/forge): ReviewCommentWriter + GitHub/Gitea implementations,
type-asserted like PullRequestReader so a forge that cannot is skipped rather
than faked (NoopForge is pinned as not satisfying it). GitHub loops one
CreateComment per comment and joins the failures, so one refused anchor cannot
drop the rest; Gitea carries them on a single COMMENT-state review.
Two deliberate departures from the PRD's sketch, both recorded in it:
- The interface is batch (CreateReviewComments), not one call per comment:
Gitea's only inline shape is one submitted review, and the set then crosses
NATS once instead of once per finding.
- The head SHA is resolved inside each forge implementation rather than by the
caller. The stage runs in archie-agent, which holds no forge credentials and
has no PullRequestReader; the daemon-side forgerpc handler is the only place
that does.
Transport (internal/forgerpc): workflow.Forger gains the method, so the agent
reaches the daemon over the new create_review_comments subject; the server
type-asserts the writer and answers an incapable forge with an error the stage
logs. This is the one proxied method whose absence on an image skew does not
park a task.
The reviewer prompt and record_finding schema ask for a suggestion on a
confirmed, mechanically fixable defect only.
Regenerating the wfextract symbol table -- the documented way a new exported
workflow symbol reaches interpreted stage plugins -- also picked up symbols that
were missing from the committed table (the Status* and Source* constants,
ReviewCheck, Store, Task, and the _Store wrapper). Nil-guards were re-applied to
every wrapper afterwards, as that file's header requires.
Evidence: task check exit 0 (incl. 229 UI tests). Three mutations of the new
stage logic each fail a test that names the behaviour -- dropping the
confirmed-only gate, dropping the Line > 0 filter, and letting a posting failure
escape -- so the tests discriminate rather than passing vacuously.
* fix(build): drop the Dockerfile pin assertion
The guard fetched golangci-lint's go.mod at image build time to fail the build
when GOLANGCI_LINT_VERSION and GOFUMPT_VERSION disagreed. It only ever tripped
on self-inflicted pin drift, but it added a network fetch and a new failure
mode to the release build, which is not where that belongs.
The pins stay correct (golangci-lint 2.13.2, whose vendored gofumpt v0.11.0
equals GOFUMPT_VERSION) and the comment above them still states the invariant,
so the reasoning survives without the build-time machinery. The pin change
itself reaches the agent image through the normal release path: CI rebuilds
and pushes archie-agent on an archie/vX.Y.Z release, and a gateway-only
release deliberately does not.
* feat(workflow): add the remediate workflow for PR review remediation
Implements step 5 of docs/prds/pr-review-remediation.md: a remediate
workflow that resumes the task's existing worktree onto its PR branch
(worktree.Manager.Resume, not Prepare's refresh), runs a builder agent
against the review's actionable comments, commits and pushes to the
same branch, and replies via the forge's ReplyToReview. Task.RetryCount
doubles as the remediation round counter, capped by the repo's
max_retries; the round-cap stage parks the task and posts one comment
when it's hit.
Task gains a ReviewPayload field (JSON-encoded review unit) with full
store/proto/staterpc plumbing. workflow.Forger gains Comment and
ReplyToReview, proxied to archie-agent via a new forgerpc RPC. Resume
and its Dir lookup are added as an unexported optional Trees capability
rather than widening the exported interface, which is already at
golangci-lint's interfacebloat cap and is projected into wfextract's
Yaegi symbol table for repository-authored stage plugins.
The daemon-side reaction consumer that dispatches into this workflow
(subscribing ARCHIE_REACTIONS, grouping comments by review, enforcing
per-task serialization) is tracked separately in archie-core-8li9.4.
* feat(messaging): extract standalone messaging service binary and process composition
* feat(messaging): adapt channel frontends to consume Gateway ChatContract
* feat(messaging): remove in-process channels and Router from daemon bootstrap
Telegram, email and webhook now run only in the Messaging Service, which
reaches this process through the Gateway's ChatContract. The daemon builds
no channel and no gateway.Router, so the wiring that existed only to serve
them goes with it: setupGateways, setupEmailGateway, setupWebhookGateway,
setupTelegramGateway, buildTelegramRouter, makeTelegramReload and the
release-announcement path helper.
The seams those paths owned move to where they can still be honoured:
- Channel ValidateConfig runs in archiemessaging.compose, which now returns
an error, so a configured-but-invalid front-end fails the service rather
than being dropped silently.
- The inbound rate limiter moves to the Gateway process, which owns the only
Router and therefore the only place a per-sender budget applies to every
channel's turns.
- The daemon's gateway readiness probe becomes a ContractProbe against the
Gateway. It cannot observe a channel lifecycle it no longer runs, and a
channel-count probe would have reported every daemon degraded.
- telegramSetup is now chatSetup (in chat_setup.go), trimmed to the fields a
chat turn runner actually reads.
Telegram operator features that were wired only in the daemon (/update,
release announcements, version reporting, SIGHUP reload) are not yet wired
in the Messaging Service: archie-core-8cda.6.6. Channel lifecycle reporting:
archie-core-8cda.6.8. The Gateway process builds no worktree manager, so
review_pr is unadvertised on every channel: archie-core-8cda.6.7.
* feat(messaging): re-wire the Telegram operator surface in the Messaging Service
setupTelegramGateway satisfied eight telegram.Gateway seams from the daemon's
own process. Each is resolved by where the fact it needs is owned, not by
where it used to be built; the decision table is in
docs/architecture/migration-decisions.md.
Wired here: /update (built from this service's own [chat.telegram] check and
install commands), the update-report relay (same sha256(bot_user)[:8] path the
daemon wrote, so a pending report is found rather than written afresh),
show_tool_calls, and /restart's config reload, which re-resolves the token and
allowlist from this service's own file and refuses a reload that resolves to
an empty token.
/version is sourced from the Gateway: ChatSnapshot gains a Version field
carrying the Gateway's rendered build block, read per call and bounded by the
dependency timeout. cmd/archie-messaging deliberately takes no version
ldflags, so it cannot report a partial upgrade as a matched one.
RunningVersions and ReleaseAnnouncements stay nil. Both need a component's
real running version, and the only versions this process can obtain are a
different binary's stamps -- reporting them would manufacture exactly the
false success the update check exists to catch, and the announcer without
parseable versions is a no-op that reads as configured. archie-core-8cda.6.9
carries the contract that unblocks them. Dangerous stays nil: no composition
has ever supplied a command authority.
Enrich is dropped from the update service: componentInstallTypeEnricher read
[nats], which this service must not decode, so the check command's own
reported install type stands.
* fix(workflow): guard inline review comments against head drift
Two findings from the review of q9au.
A comment's line numbers describe exactly one revision of the branch, but the
forge resolved the pull request's head at posting time -- after the review had
already read the worktree. A push in that window (a collaborator, a retry) meant
the findings were anchored to whatever now occupied those lines, mislabelling
instead of failing. The revision the review measured now travels with the
comment set: OpenPR records it from the same worktree read it already performs
(TaskContext.ReviewedHeadSHA), StagePostReviewComments threads it through
Forger to forgerpc, and the forge refuses the whole set once the head it reads
has moved off it -- which the best-effort stage logs, the PR body remaining the
record. An unmeasured revision posts unverified rather than dropping every
finding, so a worktree that cannot report one degrades instead of regressing.
Suggestion was documented single-line only, because GitHub applies a suggestion
fence up to its first line and a multi-line replacement would silently corrupt
the file. Only the verdict was validated, so a multi-line suggestion reached the
renderer and posted as that truncated fence. Validate now rejects an embedded
newline, which makes record_finding refuse it with feedback the model can act on
instead of the reviewer discovering the corruption later.
The events table had no attempt column, so two attempts of the same task were
indistinguishable: a retried task's stage events interleaved into one timeline and
nothing could say which run produced a given result.
- events gains an `attempt` column through the existing presence-gated
migrateTasks loop rather than a version bump, so user_version stays 3 and an
existing archie.db is migrated in place. A legacy-database test builds the
pre-change schema and proves both that existing rows survive unattributed
(attempt = 0) and that a new value round-trips.
- Event.attempt = 11 and ReadTaskLogRequest.stage = 9 are field additions to
existing messages: no new RPC, no new table, and no change to any canonical error
message that crosses the state-store wire.
- Every task event the workflow engine emits now carries the attempt. Producers
that have no attempt yet (task_queued, binding_dispatch_failure) stay
unattributed rather than guessing one.
- Changed files are captured as a durable changes_captured event at commit and push
time, reusing the existing three-dot diff primitive: paths from FilePatches(),
counts from Stats(), base_sha recorded as the merge base actually diffed rather
than the base tip, a 200-file cap with totals kept complete. Consumption goes
through an unexported optional interface so workflow.Trees -- the yaegi capability
surface for interpreted stage plugins -- is not widened. A capture is also taken
once the pull request exists, so the capture can link it.
- The per-attempt configuration is captured as a durable config_captured event
carrying the non-secret config.TaskConfig, instead of a new table plus two new
state-store RPCs: events are already this system's per-attempt provenance stream,
so a table would have given one feature two provenance mechanisms. Redaction is
proven against canary secrets. Measured payload: 539 bytes typical, 2333 bytes at
20 models.
- Captures carry a schema marker and are refused unless their key set matches the
declared payload shape, so drift surfaces as "recorded but could not be read"
rather than as a capture of zeroes.
- Three read endpoints serve the attempt rail, the changed-file captures and a raw
debug view, mapping to UI-owned response types rather than dumping store types.
- The task-log read gains a stage filter, bound to the stage logger for that
stage's own lines; agent and tool output records no stage.
A task-scoped grant may already insert events on its own task, so this provenance
is exactly as trustworthy as the rest of that task's event stream. That limit is
documented rather than papered over.
Self-review of #784: (1) the tc.ReviewReport = report stash was untested, so a
regression dropping it would silently omit the findings section; (2) the
section rendered after the workflow footer, reading as content-after-signature.
Move the findings before the footer and pin the StageReview->implementPRBody
integration with a test.
StageReview now stashes its ReviewReport on TaskContext, and implementPRBody
appends a findings section when the review ran -- explicit in the zero-findings
case ('ran and found no defects') and listing non-blocking warn/plausible
findings. A disabled or never-run review renders no section, so silence is not
mistaken for a clean review. The parked-task park reason already carries
blocking findings (renderReviewDetail), satisfying the dashboard/api-tasks half
of the acceptance criteria.
Closes archie-core-h019.6.
* refactor(workflow,store): relocate Task/Status/Source and define workflow.Store
Move store.Task, store.Status*/store.Source*, and the 3-method workflow
store interface out of internal/store into internal/domain/workflow as
workflow.Task/workflow.Status/workflow.Source plus a consumer-owned
workflow.Store, superseding store.WorkflowStore. internal/domain/workflow
now imports zero internal/store symbols; internal/store persists the
domain type instead (infrastructure depends on domain), and its Store
satisfies workflow.Store.
Updates every production and test caller of the relocated symbols across
daemon, webui, gateway, agentworker, agenttransport/nats, storerpc,
taskrun, and worktreerpc, and removes TestRecordDispatchViaExplicitTx per
docs/prds/state-store-contract.md (rev. 2c) sections 4 and 12 step 2.
Two workflow-package integration tests that exercise the real *store.Store
move to an external workflow_test package (store_integration_test.go) to
avoid a workflow(test)->store->workflow(prod) import cycle now that store
depends on workflow.
* no-mistakes(document): docs(state-store): mark §4.8 relocation done in state-store-contract
Confirmed in production (task 6, 2026-09-02): dashboard/PR summaries
showed raw prompt-token sums with no cache accounting -- 360,018 of a
reported 400,021 "prompt tokens" were cache hits billed at a steep
discount, not full price. Real fresh spend was ~40,003 tokens, roughly
10x less than the headline number implied. Nothing was double-counted;
the totals were internally consistent, just presented misleadingly.
Adds TaskContext.RunUsage (in-memory, presentation-only) and
formatTokenUsage, rendering "N tokens (F fresh + C cached)" in PR
bodies and the dashboard timeline instead of a raw sum, falling back
to the plain total when no breakdown is available.
Also trims StageBaselineGate's retry-mission prompt to failing-only
go test output via extractFailingGateOutput -- the full passing-
package listing was a direct, mechanical contributor to the growing
per-iteration prompt sizes on every gate-repair retry, not just a
display problem. Degrades to unchanged output for non-Go gates or
unrecognized shapes.
Confirmed in production (carina, 2026-09-02, task 6): StageBaselineGate
spent 405,004 tokens genuinely fixing two broken pre-existing tests and
committed the fix locally, then the plan/build stages correctly found
nothing further to do for a content-free task. StageCommitPush decided
"no changes required" purely from the build stage's own empty diff,
with zero awareness that StageBaselineGate had already committed a
real, gate-verified change two stages earlier -- so it never pushed,
never opened a PR, and the whole worktree (including the real fix) was
discarded. 669,421 tokens burned to fix a real bug that then evaporated,
and it recurs: any task that hits the same red baseline pays the
identical tax again since the fix never lands.
Adds TaskContext.BaselineFixed, set when StageBaselineGate's own
CommitAll reports a real commit. The build stage's no-changes
determination now respects it, and StageCommitPush pushes whenever
either a fresh commit was made or a baseline fix already landed,
instead of erroring on an empty tree.
Implements archie-core-h019.3 (reviewer isolation) and h019.4 (wiring),
per docs/prds/adversarial-self-review.md.
- internal/domain/workflow: new Reviewer interface and StageReview,
inserted between StageDiffCap and StageOpenPR in the implement
workflow. Fail-closed: a review that didn't run, or a confirmed
error-level finding, parks the task (store.StatusParked) with
findings in Detail; warn/plausible findings never block. Inert
unless Repo.ReviewEnabled.
- internal/worktree: Manager.Snapshot exports HEAD's tracked files into
a fresh directory with no .git, for the reviewer's isolated
workspace. Trees interface gains Snapshot; hybridTrees delegates to
it.
- internal/app/agentworker: subagentReviewer implements Reviewer using
ai-sdk's agent.Subagent (nested, in-process, non-streaming
generation) -- conversation isolation is structural (Subagent.Run
never carries message history), workspace isolation is the snapshot
above. The reviewer's toolset is read-only (read/grep/find) rooted
at the snapshot, never the worker's own toolset, so it cannot reach
the worktreerpc publication grant. Findings are captured via a typed
record_finding tool appending to a slice the stage owns, read even
when Subagent.Run errors (a truncated run's captured findings
survive).
- internal/config: Repo.ReviewEnabled (default false) and
Models["reviewer"] (falls back to Models["builder"], no schema
change).
Judgment calls: Reviewer.Review never returns a Go error -- every
failure folds into ReviewReport per the PRD's fail-closed contract,
which the stage can act on uniformly. A Subagent.Run error with zero
captured findings resolves to ReviewStatusNotRun (park); a
Subagent.Run error with >=1 captured findings resolves to
ReviewStatusCompleted with those findings, since the PRD's own example
("three defects survive a step-limit truncation") requires them to
still be evaluated for blocking rather than discarded.
Task runs now preserve real model context limits, provider usage, baseline-repair spend, commit identity, durable agent-finish telemetry, and useful timeline details without imposing a token cap. Dashboard forge links are canonical per owning GitHub or Gitea identity. Chat turn claims acquire SQLite writer intent before reading, eliminating the concurrent SQLITE_BUSY snapshot race.
Render completed ToolCallEvents as Telegram-safe streaming blocks
with compact name/status and a short cleaned preview. Summarize
file contents and listings instead of dumping source or quoted
JSON. Collapse repeated legacy turn-budget failures.
Remove max_tokens turn gating and the 200000 turn_budget_chars
default so absent or commented config cannot restore an aggregate
cap. Keep per-result max_result_chars and spill.
Confine find/grep discovery to the active workspace/worktree.
Refs #604
archie-core-467 wired tool-call visibility into the interactive chat
gateway, but a workflow task's own transcript never showed what its
tools actually did -- only stage/agent-finish events existed, so
there was no way to confirm an agent's summary matched what it really
ran.
Adds a ToolCallReporter, invoked once per completed call from
toolExecute (the single choke point every tool call passes through
before reaching a model), reported via a new ToolSetOptions.OnToolCall
hook. Runner.Run grows a report parameter to carry it in: InProcessRunner
wires it through, since it executes in the same process as its tools;
SubprocessRunner and NATSRunner accept and ignore it, since their tool
calls happen in a different process with nothing here to observe them.
The workflow engine turns each report into a tool_call event on the
task's own Bus (agent.go, implement.go), riding the same NATS
event-forwarding archie-core-518 already built to get container
events onto the daemon's timeline.
Scoped to the central tool registry (read/write/edit/grep/shell/...)
built via BuildToolSet; capture tools (structured finish output) and
run_go_script build their ToolSet entries directly and are not yet
covered.
StageBaselineGate captured a failing gate command's full output but
discarded it: a 200-char clip went into a Warn log, and the terminal
error carried only res.Status. A park reason like "go build ./...
fails ... (status: failed)" gave no way to see the actual compiler
error, and the container that had it is already gone (AutoRemove).
Now the park error carries a bounded tail of the real output (tail,
not head, since a failing test's explanation is at the end after
pages of "ok" lines), rune-safe so it can't emit invalid UTF-8 into
a persisted park reason. The three existing byte bounds in this
stage (log line, agent mission, park error) are named and documented
as independently sized rather than left as bare numbers.
Also adds docs/architecture/observability.md, the design doc for the
larger task-observability epic (#450) this fix is the first step of.