mirror of
https://github.com/multica-ai/multica.git
synced 2026-07-17 07:09:53 +02:00
agent/lambda/768b92e0
40 Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
Bohan Jiang
|
4fac8d772f |
feat(attribution): Human Attribution Phase 1 (MUL-4302) (#5150)
* feat(attribution): Phase 1 foundation — provenance schema + resolver (MUL-4302)
Human Attribution, Phase 1 (地基) first increment. Every agent run must be
traceable to exactly one accountable human AND record at which waterfall level
that human was resolved, so a NULL originator can be told apart from a genuine
'no human in the chain'.
- migration 149: add originator_source (waterfall label) + delegation/retry/
rerun/rule-version lineage + kind-tagged trigger evidence to agent_task_queue.
No FK, no cascade, no CHECK on the source enum (MUL-4302 §7); nullable ADD
COLUMNs = fast metadata-only change on the hot queue table.
- internal/attribution: the accountable-human vocabulary (Source, EvidenceKind,
TriggerKind) + pure, unit-tested classification rules (ClassifyComment/
ClassifyDirect). No DB, no authorization — provenance labeling only.
- service: attributionFor{IssueTask,TriggerComment} gather facts and delegate to
the pure classifier; the legacy originator resolvers now delegate here so
there is one source of truth. originator_user_id's VALUE is unchanged, so the
Composio-overlay and canInvokeAgent A2A authorization boundaries are
byte-for-byte preserved (MUL-4302 §1.3).
- enqueueIssueTask / enqueueMentionTask stamp originator_source + evidence;
CreateRetryTask carries the parent attribution forward and records
retry_of_task_id so retry and manual rerun stay separable (MUL-4302 §5).
Verified: go build ./..., go vet, gofmt clean; new attribution unit tests +
enqueue stamping integration test green; existing resolve_originator tests
unchanged.
Co-authored-by: multica-agent <github@multica.ai>
* feat(attribution): split accountable_user_id from originator, close enqueue bypasses (MUL-4302)
Phase 1, per Bohan's decision on the MUL-4302 thread: audit and authorization
answer different questions and get different columns.
- Migration 151 adds agent_task_queue.accountable_user_id (no FK, no cascade).
Authorization keeps reading ONLY originator_user_id (canInvokeAgent A2A gate,
Composio overlay); audit/UI/usage read accountable_user_id + source + evidence.
- Invariant (finalizeAttribution, single chokepoint + §11 tests): originator
non-null ⟹ accountable equals it. The two diverge only when originator is null
(autopilot / degraded fallback), which is the deferred rule_owner/owner_fallback
increment; this lands the column + mirror-write so that split has a home.
- Close the NULL-source enqueue bypasses Elon flagged: chat, quick-create,
deferred-fallback and run_only-autopilot now stamp originator_source + evidence
(+ accountable where a human exists). Autopilot stays unattributed until the
rule-version snapshot table lands, but is no longer a silent NULL-source row.
Retry inherits accountable_user_id like the rest of the attribution lineage.
- Fix assign/promote attribution (§4): a member who assigns/promotes an existing
issue is now the accountable human (and, by the invariant, originator) ahead of
the issue creator. Threaded as an OPTIONAL actor override, so comment/rerun/
autopilot paths keep today's resolution and create-with-assignee (creator ==
actor) is unchanged. The squad leader gate already judged the same member.
Also merges origin/main: renumbers the attribution migration 149→150 (main took
149 for issue_origin_agent_create) and folds agent_create into ClassifyDirect's
origin inheritance.
go build/vet/gofmt clean; attribution unit tests + service stamp/actor tests +
handler suite pass on a fresh DB migrated through 151.
Co-authored-by: multica-agent <github@multica.ai>
* docs(attribution): fix accountable NULL semantics + close chat/quick-create evidence boundary (MUL-4302)
Addresses Elon's 2nd-round review on PR #5150 (pre-merge doc/evidence items):
- Migration 151 no longer overclaims NULL. accountable_user_id is NULL not only
on pre-migration rows but on NEW rows whose audit source resolved no human yet
(run_only autopilot writes originator_source='unattributed' with NULL
accountable until rule_owner lands). Header + COMMENT ON COLUMN reworded so a
schema reader does not misjudge the invariant.
- Chat now uses the UNIFORM evidence pair (kind=chat, ref=chat_session_id), like
autopilot_run/issue_assignment, instead of relying only on the dedicated
chat_session_id column — new EvidenceChat kind. Added a service test asserting
chat stamps direct_human + chat evidence.
- Quick-create is documented as the ONE intentional no-antecedent-row path: no
comment/issue/session/run exists at enqueue time (the run creates the issue), so
trigger_evidence_kind/ref stay NULL while the human rides originator/accountable
and source is direct_human — not a NULL-source bypass.
No authorization behavior change. attribution + service + handler suites pass on a
DB migrated through 151.
Co-authored-by: multica-agent <github@multica.ai>
* chore(attribution): renumber migrations 150/151 → 157/158 after merging main (MUL-4302)
main's #5162 ("unblock release migrations") renumbered the chat migrations and
took 150 (agent_task_coalesced_comments) and 151 (chat_read_cursor), colliding
with this branch's attribution migrations. Renumber them above main's new highest
(156) so TestMigrationNumericPrefixesStayUniqueAfterLegacySet passes:
- 150_agent_task_attribution → 157_agent_task_attribution
- 151_agent_task_accountable_user → 158_agent_task_accountable_user
Fixed the internal "migration 150" references in 158's header to 157. Migrations
apply cleanly through 158 on a fresh DB; migration lint green.
Co-authored-by: multica-agent <github@multica.ai>
* feat(attribution): autopilot rule_owner — accountable = rule version publisher (MUL-4302)
Implements rule_owner (MUL-4302 §3.4), the first attribution source where the
accountable human diverges from the (NULL) authorization originator.
- Migration 159 adds the append-only autopilot_rule_version snapshot table (no FK,
no cascade); migration 160 adds its CONCURRENTLY lookup index.
- Write-on-publish: CreateAutopilot appends v1 (publisher = creator); UpdateAutopilot
appends a new version when a SUBSTANTIVE autopilot-row field changes (assignee /
status / execution_mode) — cosmetic edits (title/description/template) write none.
Both run inside the existing handler tx (atomic with the autopilot write).
- Dispatch resolution: both autopilot execution modes now resolve the active rule
version and stamp originator_source='rule_owner', accountable_user_id=publisher,
rule_version_id=<snapshot>, with originator_user_id left NULL (authorization
unchanged). run_only stamps CreateAutopilotTask directly; create_issue resolves in
attributionForIssueTask so both modes attribute identically. A missing version /
non-member publisher degrades to unattributed — never fabricates a human.
- finalizeAttribution now enforces the invariant ONE-WAY: it mirrors originator onto
accountable only when originator is valid, leaving an explicitly-set accountable
(rule_owner / future owner_fallback) intact when originator is NULL. Added
rule_version_id to CreateAgentTask so the create_issue path persists it too.
Also merges origin/main and renumbers this branch's attribution migrations
150/151 → 157/158 (main's #5162 took 150/151); rule_version table is 159/160.
Tests: attribution unit RuleOwner + one-way invariant table; service integration
tests proving an autopilot-origin issue stamps rule_owner + rule_version_id (and
degrades to unattributed with no version). Full service/attribution/handler/
migration suites pass on a DB migrated through 160; build/vet/gofmt clean.
Deferred (same PR): trigger-table republish (cron/webhook/event_filters) and
system-pause/archive versioning; owner_fallback + fail-closed; manual rerun.
Co-authored-by: multica-agent <github@multica.ai>
* fix(attribution): manual autopilot trigger → direct_human to the triggering member (MUL-4302)
Elon's blocking finding: a member manually triggering an autopilot was attributed
rule_owner (accountable = rule publisher, originator NULL) like a schedule/webhook
run, so member B triggering member A's autopilot landed accountable=A and carried
no originator authorization context for the run. Per MUL-4302 §4 a manual "run now"
is a direct human action and must attribute direct_human to the triggering member.
- Thread the triggering member from TriggerAutopilot into dispatch: new
DispatchAutopilotManual carries actorUserID (resolved via resolveActor +
memberActorUserID, so only a member actor is a human; an A2A agent actor falls
back to rule_owner). DispatchAutopilot / DispatchAutopilotForPlan keep their
public signatures (pass an invalid actor); only the internal dispatchAutopilot /
dispatchCreateIssue / dispatchRunOnly gained the param, so the many existing
callers are untouched.
- run_only: dispatchRunOnly stamps direct_human (originator == accountable ==
actor, no rule_version) for a manual actor, else rule_owner. CreateAutopilotTask
gains an originator_user_id param for the manual case.
- create_issue: dispatchCreateIssue enqueues a manual trigger via the actor-carrying
*WithHandoff entry points; attributionForIssueTask's autopilot-origin rule_owner
branch is now guarded on !actorUserID.Valid, so a valid actor falls through to the
direct_human override. Both execution modes attribute identically.
- schedule / webhook keep rule_owner (no actor). Trigger-table + system-pause/archive
versioning remain the pre-merge follow-ups.
Tests: the run_only row assertion Elon asked for (schedule → rule_owner row on
CreateAutopilotTask), plus manual direct_human on BOTH modes (run_only and
create_issue), including a manual actor distinct from the rule publisher. Full
service/attribution/handler/migration/scheduler/cmd suites pass on a DB migrated
through 160; build/vet/gofmt clean.
Co-authored-by: multica-agent <github@multica.ai>
* feat(attribution): owner_fallback + fail-closed policy, and manual-rerun direct_human (MUL-4302)
Two of the three remaining Phase 1 items (trigger-table / system-pause versioning
is deferred — see PR description).
owner_fallback + fail-closed (§1/§3.5) — the never-null accountable guarantee:
- attribution.OwnerFallback degrades an UNATTRIBUTED result to owner_fallback:
accountable = agent owner, originator stays NULL (audit-only, authz untouched),
Source.Precise()==false. finalizeAttribution's one-way invariant already allows
accountable-set / originator-NULL divergence, so nothing else changes.
- Migration 161 adds workspace.attribution_fail_closed (default FALSE) + a lean
GetWorkspaceAttributionFailClosed read. (Also added the column to ListWorkspaces'
explicit column list so its row type stays db.Workspace.)
- applyAttributionFallback is applied at every enqueue boundary (issue, mention,
chat, quick-create, deferred-fallback, autopilot run_only): unattributed →
owner_fallback (agent owner) by default, or ErrAttributionFailClosed when the
workspace is fail-closed, which the caller surfaces to refuse the enqueue (the
run does not start). So no run is left without an accountable human, and a
compliance workspace can block unattributable runs instead.
manual rerun (§5) — a rerun is a NEW direct_human trigger to the rerunning member:
- RerunIssue threads the acting member (resolved in the handler via resolveActor)
down to enqueueRerunTask, and attributionForIssueTask is now actor-first so the
actor wins over an INHERITED trigger comment (a rerun keeps the comment for the
daemon's prompt context but must attribute to whoever clicked rerun, not the
original comment's human).
- rerun_of_task_id lineage is recorded via a targeted SetAgentTaskRerunOf update on
the rerun path only (keeping the shared CreateAgentTask insert untouched), so
system retry (retry_of_task_id) and human rerun stay separable in reporting.
Tests: OwnerFallback unit test; owner_fallback + fail-closed-refusal + manual-rerun
(direct_human + rerun_of_task_id) service tests; the prior "degrades to unattributed"
test updated to owner_fallback. Full service/attribution/handler/migration/scheduler/
cmd suites pass on a DB migrated through 161; build/vet/gofmt clean.
Co-authored-by: multica-agent <github@multica.ai>
* fix(attribution): close fail-open holes + move rerun_of_task_id into creation snapshot (MUL-4302)
Addresses Elon's two must-fixes on PR #5150.
1. accountable-never-null / fail-closed had fail-open holes. applyAttributionFallback
now, for an UNATTRIBUTED run, refuses the enqueue (ErrAttributionFailClosed) in
THREE cases instead of silently degrading to a runnable NULL-accountable task:
- workspace policy read fails (or no workspace) → fail closed; we cannot confirm
fallback is permitted, so we don't run an unattributable task on a DB hiccup.
(Only the rare unattributed path pays this; precise runs never read the policy.)
- workspace is fail-closed → refuse (unchanged).
- owner_fallback has no valid agent owner → refuse rather than enqueue a task with
a NULL accountable_user_id.
ErrAttributionFailClosed's doc now covers all three "cannot guarantee an
accountable human" refusals. Added missing-owner / policy-read-failure /
precise-passthrough tests.
2. manual rerun rerun_of_task_id was a post-notify UPDATE (race: the queued event /
daemon claim could see rerun_of_task_id = NULL, and a failed update degraded the
run to a plain direct_human). It now rides the CreateAgentTask insert — threaded
through enqueueIssueTask / enqueueMentionTask as a creation param (like
retry_of_task_id) so it is written in the same statement before the daemon is
notified. Removed the SetAgentTaskRerunOf follow-up query.
Also merges origin/main (unrelated CLI fix #5167, no conflict). Full service /
attribution / handler / migration / scheduler / cmd suites pass on a DB migrated
through 161; build / vet / gofmt clean.
Co-authored-by: multica-agent <github@multica.ai>
* feat(attribution): rule_owner versioning on trigger edits + system-pause/archive (MUL-4302)
The final remaining Phase 1 item: substantive publishes beyond the autopilot row now
republish the rule version, so a run's rule_owner accountable follows whoever last
changed what the rule does.
- Extracted the config-summary + insert into service.RecordAutopilotRuleVersion so
the handler and the (different-package) failure monitor share one writer; the
handler's recordAutopilotRuleVersion is now a thin wrapper.
- Trigger edits: UpdateAutopilotTrigger and DeleteAutopilotTrigger republish the rule
version with the acting member as publisher, ATOMICALLY (tx-wrapped mutation +
version write, mirroring CreateAutopilot/UpdateAutopilot). CreateAutopilotTrigger
republishes best-effort — the webhook path mints its token with a retry loop that
cannot share one tx, and a create is usually initial setup already covered by v1;
a failed write there is benign (active version stays the current publisher, the new
trigger fires under it, no immediate daemon claim rides it).
- Archive (DeleteAutopilot) republishes (member, status=archived), tx-wrapped.
- System auto-pause (failure monitor) republishes with a 'system' publisher,
best-effort — a background sweep to a non-dispatching state (a paused autopilot
never dispatches; a later member resume supersedes).
- RotateWebhookToken / SetSigningSecret deliberately do NOT version: they rotate
credentials, not the rule's behavior (not §3.4 substantive).
Semantics: a system-published (no-member) active version degrades dispatch to
unattributed → owner_fallback, never fabricating a human.
Tests: republish-reattributes (member A → member B supersedes → dispatch resolves to
B; system publisher → unattributed). Full service/attribution/handler/migration/
scheduler/cmd suites pass on a DB migrated through 161; build/vet/gofmt clean.
Also merges origin/main (unrelated frontend feature #5074).
Co-authored-by: multica-agent <github@multica.ai>
* fix(attribution): make trigger-create rule-version republish atomic (MUL-4302)
Addresses Elon's final Phase 1 blocking finding: CreateAutopilotTrigger recorded the
rule-version republish best-effort AFTER the trigger insert. If member B added a
schedule/webhook trigger to member A's autopilot and the version write failed, future
schedule/webhook dispatches would keep attributing to A — violating the rule_owner
invariant that the last member to substantively change the rule owns future runs
("no immediate daemon claim" doesn't save it, since the miss surfaces at the LATER
trigger firing).
Both create paths now write the version in the SAME tx as the trigger INSERT:
- schedule create: wrap CreateAutopilotTrigger + recordAutopilotRuleVersion in one tx.
- webhook create: each mint-with-retry attempt runs in its own tx (insert + version
commit together; a token collision rolls that attempt back and retries with a fresh
token; a version-write failure rolls the trigger back). Passes ap + the acting
member id into the helper.
- removed the best-effort recordTriggerRuleVersionBestEffort helper (and the now-unused
slog import).
Test: TestCreateTrigger_RepublishesRuleVersionAtomically drives both create paths
through the handler and asserts a rule version is published by the acting member.
Existing webhook/trigger/archive handler tests still pass. Also merges origin/main
(unrelated avatar feature #5074). Full service/attribution/handler/migration/
scheduler/cmd suites pass on a DB migrated through 161; build/vet/gofmt clean.
Co-authored-by: multica-agent <github@multica.ai>
* feat(attribution): Phase 2.1 — surface run attribution on the task API (MUL-4302 §9)
First Phase 2 (visibility) increment: the agent-task API now returns the resolved
accountable-human provenance so the UI can render an "on behalf of" badge.
- AgentTaskResponse gains an `attribution` object: source label (never blank —
pre-migration NULL renders "unattributed") + `precise` flag (false for the degraded
owner_fallback / backfill / unattributed sources), the initiator (accountable) and
originator (authorization) user refs, the evidence {kind, ref_id} pointer, and the
rule_version / delegated / retry / rerun lineage ids.
- The label + evidence + raw ids are built in the PURE taskToResponse (no DB), so
every task response carries them. Names are hydrated separately, only on the
user-facing surfaces (ListAgentTasks, ListWorkspaceAgentTaskSnapshot, RerunIssue,
CancelTaskByUser) — daemon-claim paths stay lean.
- Hydration resolves initiator/originator from the GLOBAL user table (departed-member
safe) via a new batch GetUsersByIDs query (no N+1); best-effort, so a lookup hiccup
leaves the raw ids intact.
Tests: pure taskAttributionBase (direct_human / rule_owner NULL-originator /
owner_fallback degraded / pre-migration→unattributed) + DB hydration (fills known
ref, leaves unknown id un-filled, skips nil). Full handler/service/attribution/
migration/scheduler/cmd suites pass on a DB migrated through 161; build/vet/gofmt
clean. The field is additive — the frontend's parseWithFallback ignores unknown keys,
so nothing breaks until the UI increment consumes it.
Also merges origin/main (unrelated editor feature #5090).
Remaining Phase 2 (next increments, same PR): frontend zod schema + "on behalf of"
badge + evidence-chain jump; append-only correction events (write + display).
Co-authored-by: multica-agent <github@multica.ai>
* feat(attribution): Phase 2.2 — on-behalf-of badge in the execution log (MUL-4302 §9)
Surface the accountable human on every agent run row:
- AttributionBadge composes Badge + ActorAvatar, shows "on behalf of <member>"
with the resolution source as a tooltip; degraded (non-precise) attribution
gets a warning tone, and an unresolved initiator renders an explicit
"no responsible member" chip.
- Wire the badge into both active and past rows of the execution log.
- Mirror the attribution shape into AgentTaskResponseSchema (defensive, .loose())
so the cancel-task path carries it through zod; add parse tests.
- Export TaskAttribution/AttributionUser/TaskEvidence from @multica/core/types
and add the attribution block to all four issues.json locales.
Co-authored-by: multica-agent <github@multica.ai>
* fix(attribution): hydrate initiator names on issue-facing task endpoints + bound the badge (MUL-4302 §9)
Address Elon's PR #5150 review:
- ListTasksByIssue (the execution-log data source), GetActiveTaskForIssue and the
issue-scoped CancelTask now call hydrateTaskAttributions, so the "on behalf of
<member>" badge shows the real member name on issue detail instead of falling
back to "someone". Mirrors the existing ListAgentTasks / snapshot behavior.
- AttributionBadge: cap width (max-w-40, min-w-0) and truncate the name span so a
long name / narrow right column can't squeeze out trigger/status/actions; keep
the avatar shrink-0.
- Add a handler test asserting the issue task list returns a hydrated
attribution.initiator.name.
Co-authored-by: multica-agent <github@multica.ai>
* fix(attribution): use semantic AvatarSize 'xs' for the badge avatar
main refactored ActorAvatar.size from a raw pixel number to the semantic
AvatarSize union (packages/ui/lib/avatar-size). Switch the on-behalf-of badge
avatar from size={14} to size="xs" (16px) after merging main.
Co-authored-by: multica-agent <github@multica.ai>
* fix(attribution): stage-cascade falls back to parent-issue provenance, not agent owner (MUL-4302)
When closing the last sub-issue in a Stage wakes the parent's assignee agent, the
run was enqueued via a system-authored child-done comment with no actor, which the
resolver classified as unattributed and then degraded to owner_fallback (the agent's
own owner). That is the wrong accountable human: the woken run should be accountable
to whoever caused the parent issue to exist.
attributionForIssueTask now detects a system-authored trigger comment and falls
through to the parent issue's own provenance — the same creator / agent_create-origin
/ autopilot-origin chain a direct enqueue resolves (so an agent-decomposed parent
attributes via delegation to the human who drove it; a member-created parent to that
member; an autopilot parent to the rule publisher). owner_fallback is now only the
last resort when the parent provenance itself has no human.
- Extract attributionFromComment so attributionForIssueTask can inspect author_type
without a second GetComment; authorization resolution stays byte-identical.
- Add a DB-backed test asserting a system child-done comment resolves to the parent
issue's origin human (delegation), not owner_fallback.
Co-authored-by: multica-agent <github@multica.ai>
* feat(attribution): autopilot runs attribute to the firing trigger's creator (MUL-4302)
Per Bohan: an autopilot schedule/webhook run should be accountable to the human
who created the SPECIFIC trigger that fired it, not the rule publisher. (Manual
triggers already attribute to the invoking member via direct_human — unchanged.)
- Migration 162: add autopilot_trigger.created_by_type/created_by_id (nullable, no
FK/cascade). Capture the creating member at both trigger-create sites (schedule +
webhook).
- New precise source trigger_owner: originator stays NULL (an autonomous fire
carries no human authorization — same authz-safe divergence as rule_owner),
accountable = the trigger's member creator.
- triggerOwnerAttribution resolves run.trigger_id → creator; wired into run_only
dispatch and the create_issue path (bridging issue → active run → trigger_id).
Legacy triggers with no recorded creator, and agent-created triggers, degrade to
rule_owner then owner_fallback — nothing regresses.
- Frontend: trigger_owner source label in all four locales + badge switch case.
- Tests: attribution TriggerOwner unit + Precise/invariant; DB-backed resolver
tests (member creator → trigger_owner; creatorless → rule_owner fallback).
Co-authored-by: multica-agent <github@multica.ai>
* chore(attribution): re-trigger CI (dropped synchronize event on
|
||
|
YYClaw
|
9eddcaff10 |
fix(chat): defer cancellation-time finalization until the task transcript is stable (#5246)
A quick Stop before the agent's first token no longer races a late reply. Started-but-empty cancellations defer the empty/non-empty judgment until the daemon acks its transcript flush (or a grace-period sweeper fires), then settle to a single outcome. Empty outcomes persist a durable, creator-authorized draft restore (fetched/consumed via a dedicated endpoint, reconnect-safe and at-most-once) instead of broadcasting the prompt over the workspace bus. Closes #5219 |
||
|
Multica Eve
|
e07b5403ab |
MUL-4502: make autopilot webhook admission durable (#5386)
* fix(autopilots): make webhook admission durable Co-authored-by: multica-agent <github@multica.ai> * fix(autopilots): address webhook delivery review Co-authored-by: multica-agent <github@multica.ai> --------- Co-authored-by: Eve <eve@multica-ai.local> Co-authored-by: multica-agent <github@multica.ai> |
||
|
Bohan Jiang
|
cb87dd106b |
feat(chat): task-owned direct-chat input batches + explicit no_response outcome (MUL-4351) (#5195)
* feat(chat): task-owned direct-chat input batches + explicit no_response outcome (MUL-4351) Direct (web/mobile) chat no longer uses the last-assistant-row as an implicit input cursor. Each direct send now owns an immutable input batch: - agent_task_queue.chat_input_task_id makes a task the owner of the user messages it must consume; the send path creates the task + user message + attachment bindings + session touch in one transaction, and the daemon is notified only after commit. A claim reads exactly that batch, so a message that arrives mid-run belongs to the next task and is never absorbed. - Auto-retry inherits the root input owner and is queued at a bumped priority, created inside FailTask's transaction so no newer chat task can jump ahead. - CompleteTask writes exactly one assistant outcome inside the completion transaction: a normal message, or a visible no_response outcome (with a non-empty English fallback) when the final output is empty. The write failing rolls the completion back and the handler returns 5xx so the daemon retries; the status CAS keeps it idempotent. chat:done carries message_kind. - Web/desktop/mobile render no_response as a localized 'no text reply' state (keeping the tool timeline), suppress Copy, keep it unread, and keep the session-list preview non-blank. - Legacy/channel tasks (chat_input_task_id NULL) keep the trailing-message selector, so a rolling deploy never replays Slack/Lark history. Co-authored-by: multica-agent <github@multica.ai> * fix(chat): scope no_response to direct tasks; don't cancel task on input read error (MUL-4351) Addresses PR review (Niko): - writeChatCompletionOutcome only writes a no_response row for task-owned direct tasks (chat_input_task_id set). Legacy/channel (Slack/Lark) tasks keep the prior behavior: empty output writes no assistant row, so chat:done carries empty content and the channel outbound silently drops it — the no_response fallback body never reaches an external channel. - The daemon claim distinguishes a genuine zero-input batch from a failed input read: on ListChatInputMessages / ListChatMessages error it returns 5xx and preserves the dispatched task for redelivery instead of cancelling a valid task on a transient DB error. Co-authored-by: multica-agent <github@multica.ai> --------- Co-authored-by: J <j@multica.ai> Co-authored-by: multica-agent <github@multica.ai> |
||
|
Multica Eve
|
bf161f2f9c |
fix(tasks): preserve merged comment delivery (#5192)
Track actual claim-time delivery, support legacy daemons, and repair comment batches across claim, retry, edit, and delete races. MUL-4348 Co-authored-by: Eve <eve@multica-ai.local> Co-authored-by: multica-agent <github@multica.ai> |
||
|
Multica Eve
|
75695a2e40 |
fix(comments): guarantee at-least-once processing of user comments (MUL-4195) (#5068)
* fix(comments): guarantee at-least-once processing of user comments (MUL-4195) Consecutive comments on an issue were silently dropped: a new comment that arrived while the agent already had a queued/dispatched task was discarded by the HasPendingTaskForIssueAndAgent dedup, losing the user's follow-up instruction with no visible trace. Comments — unlike chat — are deliberate, addressed, persisted input and must never vanish. This makes comment handling at-least-once while keeping concurrency bounded to one run per (issue, agent): - Merge, don't drop (PR1): a comment landing while a not-yet-started task exists is folded into that task — the prior trigger becomes a coalesced comment and the new one becomes the trigger, so a single run still covers every deliberate comment. Falls back to a fresh enqueue if the pending task was claimed mid-flight, so nothing is lost in the race. - Completion reconciliation (PR2): on task completion, a member comment newer than the run's started_at schedules exactly one follow-up via the normal trigger pipeline. Loop-safe: member-authored only, capped by the existing per-(issue,agent) dedup, and terminating. - Visibility (PR3): coalesced_comment_ids is surfaced on the task API and in the run prompt so the covered comments are explicit. Migration 145 adds agent_task_queue.coalesced_comment_ids UUID[]. Tests: merge-not-drop preserves all three of a rapid burst and repoints the trigger to the newest; reconciliation query gates on member/since; e2e CompleteTask enqueues a follow-up for a mid-run member comment and does not for none. Co-authored-by: multica-agent <github@multica.ai> * fix(comments): address review — originator gate, agent-scoped reconcile, cross-thread coalesced prompt (MUL-4195) Resolves GPT-Boy's Request-changes review on PR #5068. Must-fix #1 — merge no longer inherits a stale originator/runtime context. MergeCommentIntoPendingTask now only folds a comment into a pending task whose originator_user_id IS NOT DISTINCT FROM the new comment's originator. runtime_mcp_overlay / runtime_connected_apps are a pure function of (originator, agent) and the agent is fixed, so a matching originator keeps the stored overlay/attribution valid; a differing originator (e.g. user B commenting on a task originated by user A) matches no row and the caller enqueues a fresh follow-up with B's own context instead of reusing A's. trigger_summary is refreshed to the new trigger comment. Must-fix #2 — completion reconcile no longer re-wakes unrelated agents. reconcileCommentsOnCompletion computes the latest member comment's triggers and keeps ONLY the agent that just completed, instead of fanning the comment out through the full pipeline. An @-mention of agent B during agent A's run is triggered once at creation time and is no longer replayed (double-run) when A completes. Should-fix #3 — coalesced-comment prompt no longer assumes a single thread. The claim response now carries each folded comment's thread id / author / created_at / content (CoalescedCommentData); the prompt embeds them directly so the agent addresses cross-thread folded comments without the wrong "they are in the triggering thread" hint. Old servers that ship only ids fall back to an issue-wide fetch, still without the same-thread assumption. Tests: TestMergeCommentIntoPendingTask_OriginatorGate (query gate), TestCompleteTask_DoesNotReTriggerOtherAgentMentionedDuringRun (reconcile scoping), TestBuildCommentPromptCoalescedCrossThread / IDsOnlyFallback (prompt). Existing MUL-4195 suites still pass. Co-authored-by: multica-agent <github@multica.ai> * fix(comments): close unique-index drop + dispatched-window race in comment coalescing (MUL-4195) Second-round review follow-up on PR #5068. Must-fix #1 — originator-mismatch no longer drops the comment. The previous originator gate returned ErrNoRows on a different originator and the caller fell through to a fresh enqueue, which collided with the idx_one_pending_task_per_issue_agent unique index (one queued/dispatched task per (issue, agent)) — silently dropping the second user's comment. Replaced the gate with recompute-on-merge: MergeCommentIntoPendingTask now re-stamps originator_user_id, runtime_mcp_overlay, runtime_connected_apps and trigger_summary to the new comment's originator. A different member's comment folds into the single coalescing run carrying the latest instruction's own identity/overlay (no cross-user capability bleed, no drop, no collision). Must-fix #2 — comment arriving in the claim→StartTask window is no longer lost. Merge now targets only PRE-CLAIM states ('queued','deferred'); a dispatched/running task is never a merge target, so a post-claim comment is never falsely stamped into coalesced_comment_ids as "delivered". Completion reconcile is re-anchored on dispatched_at (the moment the claim response is built) instead of started_at, and sweeps ALL undelivered member comments since that anchor — replaying each through the normal enqueue path so they coalesce into one bounded, agent-scoped follow-up run. This covers the dispatch→start window a started_at anchor missed. Enqueue path: on a merge miss the caller no longer blindly fresh-enqueues (which could collide with a dispatched sibling); it defers to the active task's completion reconcile via HasActiveTaskForIssueAndAgent, and only fresh-enqueues when no active task exists. Tests: rewrote the query test to TestMergeCommentIntoPendingTask_RecomputesOriginatorAndSkipsDispatched; added TestConsecutiveCommentsDifferentOriginatorsFullEnqueuePath (full handler enqueue path, two distinct originators) and TestCompleteTask_ReconcilesDispatchedWindowComment (claim→start window). All existing MUL-4195 handler/cmd-server/daemon/service suites still pass. Co-authored-by: multica-agent <github@multica.ai> * fix(comments): catch pre-dispatch merge-race comment in completion reconcile (MUL-4195) Third-round review follow-up on PR #5068. Race: a member comment is created while the task is still queued, but its merge loses the race to the daemon claiming the task (queued→dispatched). The merge then finds no pre-claim row (ErrNoRows), the enqueue path defers to reconcile — but the comment's created_at is BEFORE dispatched_at, so the dispatched_at-anchored reconcile skipped it and the comment vanished with no task coverage. Fix: anchor completion reconcile on the task's created_at (which always precedes dispatch) instead of a dispatch/start timestamp, and exclude the run's DELIVERED SET — trigger_comment_id ∪ coalesced_comment_ids. Because merges only ever touch pre-claim rows, that set is exactly what the claim response carried, so any member comment created since the task was made that is NOT in it was genuinely undelivered and earns a bounded follow-up. This catches the pre-dispatch merge-race comment and the dispatch→start comment, while never re-firing a comment that was delivered as a pre-claim coalesced entry. Test: TestCompleteTask_ReconcilesPreDispatchMergeRaceComment reproduces the race (comment created pre-dispatch, task dispatched before merge, plus a delivered coalesced comment) and asserts exactly one follow-up, triggered by the race comment, with the delivered coalesced comment excluded. Existing reconcile fixtures updated to set a realistic created_at (the production invariant that created_at is the earliest task timestamp). Co-authored-by: multica-agent <github@multica.ai> * fix(comments): merge only into the queued task, never a deferred fallback (MUL-4195) Fourth-round review follow-up on PR #5068. MergeCommentIntoPendingTask targeted status IN ('queued','deferred') ordered by created_at DESC. When a (issue, agent) pair had both an older queued task (the run about to be claimed) and a newer deferred assignee-fallback task, a new comment merged into the deferred row instead of the queued one — so the comment missed the imminent run and the deferred fallback could later promote into a duplicate/conflicting run. This merge is only ever reached when HasPendingTaskForIssueAndAgent matched a queued/dispatched task (it never inspects deferred), so the coalescing target must be the queued row. Restricted the merge target to status = 'queued' (the unique index guarantees at most one). Deferred fallbacks keep their own fire_at/promotion escalation lifecycle and are never a merge target. Test: TestMergeCommentIntoPendingTask_TargetsQueuedNotDeferred seeds an older queued task + a newer deferred fallback for the same (issue, agent), merges a new comment, and asserts it lands on the queued task (trigger repointed, old trigger coalesced) while the deferred fallback is left untouched. Co-authored-by: multica-agent <github@multica.ai> --------- Co-authored-by: Eve <eve@multica-ai.local> Co-authored-by: multica-agent <github@multica.ai> |
||
|
Multica Eve
|
c4997af4d1 |
fix: archive autopilots on delete (#5042)
Co-authored-by: Eve <eve@multica-ai.local> Co-authored-by: multica-agent <github@multica.ai> |
||
|
LinYushen
|
cb68669c73 |
feat(composio): gate MCP apps behind feature flag (#4876)
* feat(composio): server-side connect flow + connections REST (Notion MVP) (MUL-3720) (#4608)
* feat(composio): server-side connect flow + connections REST (Notion MVP) (MUL-3720)
Compose the merged server/pkg/composio SDK into a user-facing connection
manager: signed-state connect handshake, local user_composio_connection
mirror, idempotent disconnect, and a per-user MCP session helper (not yet
wired into task dispatch).
- migration 127_user_composio_connection (no FK/cascade, per DB rules)
- sqlc queries: upsert (idempotent on user_id+connected_account_id), list
active, owner-scoped get, mark revoked
- internal/integrations/composio: signed HMAC-SHA256 state, BeginConnect,
CompleteCallback (idempotent upsert), ListConnections, Disconnect
(upstream 404 = idempotent success), CreateMCPSession (no-op when empty,
pins connected_accounts per toolkit), CallbackRedirect
- REST handlers under /api/integrations/composio (user-scoped, 503 when
COMPOSIO_API_KEY unset): connect/init, callback (302), connections list,
delete
- router wiring gated by COMPOSIO_API_KEY; COMPOSIO_AUTH_CONFIGS_JSON maps
toolkit->auth_config (MVP: notion); state secret from COMPOSIO_STATE_SECRET
or derived from JWT_SECRET; callback base from COMPOSIO_CALLBACK_BASE_URL
or MULTICA_PUBLIC_URL
- tests: state (expire/tamper/wrong-secret), service (mapping, callback
idempotency, non-success, disconnect owner/404 idempotency, MCP pin),
handlers (httptest), redact regression for Bearer mcp_ tokens
MVP scope: Notion only; no task-dispatch overlay, sharing, or webhook
event handling (later stages).
Co-authored-by: multica-agent <github@multica.ai>
* fix(composio): bind callback account to user + idempotent revoked disconnect (MUL-3720)
Address PR 4608 review (CHANGES_REQUESTED):
- callback: verify connected_account_id with Composio before mirroring it.
The signed state only proved user/toolkit/exp, so a valid state paired with
a tampered connected_account_id would be written verbatim. CompleteCallback
now calls ListConnectedAccounts and fails closed (ErrAccountVerification)
unless the account belongs to the state's user (composio_user_id == multica
user id) and was created under the toolkit's auth config. No row is written
on mismatch / unknown account / upstream error.
- disconnect: short-circuit to a no-op when the local row is already revoked,
before touching upstream. Previously a second DELETE re-hit Composio and a
non-404 upstream error surfaced as a 502, breaking the 204-idempotent
contract.
- CreateMCPSession: document the v1 single-active-connection-per-(user,toolkit)
constraint and make duplicate selection deterministic (newest-wins, rows are
connected_at DESC) instead of order-dependent map overwrite. Stage 3 owns the
real single-account-enforcement vs multi-account-shape decision.
Tests: tampered/wrong-auth-config/unknown-account callback rejection, revoked-row
disconnect no-op (asserts upstream not re-hit). composio pkg 85% coverage; all
green.
Co-authored-by: multica-agent <github@multica.ai>
* feat(composio): list all toolkits + dynamic auth-config resolution (MUL-3720)
Yushen's follow-up to the Notion MVP: surface the full Composio toolkit
catalog, render it in Settings, and drop the static env mapping in favor of
dynamic auth-config discovery.
Config correctness (per Composio docs):
- Remove COMPOSIO_AUTH_CONFIGS_JSON entirely. The toolkit→auth_config mapping
is now resolved at request time from the project's /auth_configs (cached,
5-min TTL), so enabling a toolkit is a dashboard action, not a redeploy.
- Do NOT add COMPOSIO_PROJECT_ID. The project API key (x-api-key) authenticates
to exactly one project; the project is resolved from the key. Only org-level
endpoints use x-org-api-key, which this integration never calls.
Backend:
- SDK: server/pkg/composio/auth_configs.go — ListAuthConfigs (toolkit_slug,
is_composio_managed, show_disabled, limit, cursor).
- service: dynamic resolver (authConfigMap cache; betterAuthConfig prefers a
custom/white-label config over Composio-managed, newest wins); BeginConnect
and CompleteCallback resolve via it; ListToolkits fetches the full catalog
(paginated, capped) annotated with connectable = has an enabled auth config,
connectable-first ordering.
- handler + route: GET /api/integrations/composio/toolkits (user-scoped, 503
when COMPOSIO_API_KEY unset) returning slug/name/logo/category/connectable.
Frontend:
- core: ComposioToolkit/ComposioConnection types, api client methods, and
composio query options (@multica/core/composio).
- views: Settings → Integrations now has a Composio section rendering every
toolkit as a card with search. Connect is gated on `connectable`;
non-connectable toolkits show a muted "not configured" hint instead of a
dead button. Connected toolkits show a badge + Disconnect (with confirm).
- i18n: composio block added to en/zh-Hans/ja/ko settings.
Tests: SDK + service (dynamic resolution, custom-over-managed preference,
connectable flag, resolver-error soft-degrade) and handler toolkits endpoint;
composio pkg 85.7% coverage. go build/vet/gofmt clean; core+views typecheck,
core+views lint, and core tests (691) all green.
Co-authored-by: multica-agent <github@multica.ai>
* fix(composio): close cross-toolkit callback fail-open by signing auth_config_id into state (MUL-3720)
Re-review blocker: CompleteCallback resolved the toolkit's auth config at
callback time and ignored a resolve error/empty result, while
verifyAccountOwnership skipped the auth-config comparison when the expected
value was empty. A user could then pass another toolkit's connected_account_id
into this toolkit's callback — the owner check passed and it was written under
the wrong toolkit_slug/account binding.
Fix: the auth_config_id is already resolved in BeginConnect (before the state
is signed), so sign it into the state and compare it exactly at callback. No
re-resolve, no fail-open. verifyAccountOwnership now fails closed when the
expected auth config is empty (rejects instead of skipping) and requires an
exact match — closing the cross-toolkit binding gap.
Tests: state round-trips auth_config_id; BeginConnect signs it; callback
rejects wrong/cross-toolkit auth config and an empty (no-mapping) auth config
fails closed. composio pkg 85.2% coverage, all green.
Frontend (non-blocking): the Composio settings tab now surfaces an error when
the connections query fails instead of silently rendering everything as
unconnected.
Co-authored-by: multica-agent <github@multica.ai>
* fix(composio): hide Settings section entirely when integration unconfigured (MUL-3720)
Decision (option 2, hide-then-merge): don't show a card that leaks the internal
COMPOSIO_API_KEY env-var name to every end user. IntegrationsTab now gates the
whole Composio section (heading + body) on the toolkits query — a 503 means the
key is unset, so the section is withheld instead of rendering the not-configured
card. Admin-only setup guidance is a later, role-gated affordance.
Removed the notConfigured card (and now-unused ApiError import) from
ComposioTab; it only mounts when configured. views typecheck + lint clean.
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
* feat(composio): Stage 2 frontend polish — callback toast, last_used & expired UI, e2e (MUL-3718) (#4688)
* feat(composio): callback toast + refresh, last_used & expired UI, e2e (MUL-3718)
Co-authored-by: multica-agent <github@multica.ai>
* fix(composio): real callback redirect route + StrictMode-safe toast dedup (MUL-3718 review)
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
* fix(composio): callback endpoint should not require Multica auth (MUL-3843) (#4709)
* fix(composio): move OAuth callback out of the Auth group (MUL-3843)
Composio 302-redirects the browser to /api/integrations/composio/callback
at the end of the OAuth flow, but PR #4608 mounted it inside the cookie-auth
middleware group. When the session cookie is absent (expired session,
SameSite=Strict / Safari ITP, private window, self-hosted callback subdomain)
the Auth middleware returned a hard 401 and a JSON blob instead of the
settings redirect, breaking the flow.
Identity never came from the cookie anyway: it is carried by the HMAC-signed
state param that CompleteCallback verifies (signature, expiry, replay) and
cross-checked by verifyAccountOwnership; h.Composio == nil still 503s. So the
callback is registered alongside the other public OAuth/webhook routes; the
other four composio endpoints stay session-gated.
Refs MUL-3843, MUL-3715.
Co-authored-by: multica-agent <github@multica.ai>
* fix(composio): correct stale callback routing comments (MUL-3843)
The package header and ComposioCallback doc comments still described the
callback as sitting under the Auth middleware group. After the route was
moved out (this PR), update both to state it is a public route whose identity
comes from the signed state — addressing review nit from 张大彪.
Refs MUL-3843.
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
* feat(composio): inject MCP overlay into agent runtime at task dispatch (MUL-3721) (#4704)
Stage 3 of the Composio epic. Wires the per-user Composio MCP session into
every agent task so the agent process sees the initiator's connected tools
without any prompt-time plumbing.
Server side
- Migration 128 adds agent_task_queue.runtime_mcp_overlay JSONB plus a
BEFORE-UPDATE trigger that wipes the column on any transition into a
terminal status (completed / failed / cancelled). A trigger is the single
source of truth — future queries that flip status cannot bypass it.
- composio.Service.BuildTaskOverlay(userID) reuses CreateMCPSession and
emits the Claude-style { mcpServers: { composio: { type: http, url,
headers } } } shape the daemon's existing sidecar generators consume.
Returns (nil, nil) on zero active connections so we never burn a
Composio session for a user with nothing to call.
- TaskService grows a Composio ComposioOverlayBuilder seam, wired in
router.go after composiointeg.NewService succeeds. Five enqueue paths
(issue / mention / quick-create / chat / auto-retry) attach the overlay
after CreateAgentTask returns and before the daemon is notified — so
every claim reads a settled row, with no second daemon hop. Best-effort:
a builder failure logs and proceeds with no overlay.
- resolveInitiatorFromTriggerComment derives the initiator user from the
trigger comment when it was authored by a member. Agent-authored
triggers are not treated as initiators (their connected-apps view is
empty by construction).
Daemon side
- handler/daemon.go claim path merges task.runtime_mcp_overlay onto
agent.mcp_config via mergeMCPOverlay before populating
TaskAgentData.McpConfig. Overlay wins on server-name collisions
because it carries the live user-scoped session URL. Errors fall back
to the agent config unchanged — a bad overlay must not surprise-disable
saved MCP tools. The existing execenv sidecar generators (cursor /
codex / openclaw / opencode / hermes-kiro) need no changes: they keep
consuming the merged result through TaskAgentData.McpConfig.
Tests
- 9 merge cases (mcp_overlay_test): both-nil short-circuit, agent-only
pass-through, overlay-only canonicalization, two-side merge, name
collision (overlay wins), top-level key preservation, malformed agent
fallback, malformed overlay fallback, non-object server rejection.
- 4 dispatch cases (composio): zero-connections returns nil without
CreateSession, happy-path emits the right shape with the right user
id, empty-URL defensive branch, SDK error surfacing.
- 4 TaskService helper cases: nil Composio is a no-op (Queries-safe),
invalid initiator does not call the builder, nil overlay skips the
UPDATE, builder error swallowed without panic.
- Migration 128 verified to roll up + down + up cleanly against the test
database.
Out of scope (deferred): assignment-triggered enqueue paths with no
trigger comment get no overlay attached today (no initiator UUID flows
through enqueueIssueTask in that case). Retry paths recompute the overlay
fresh from the parent's initiator_user_id instead of inheriting the bearer
from the parent row, so a stale token can never resurface on a retry.
Co-authored-by: Eve <eve@multica.ai>
Co-authored-by: multica-agent <github@multica.ai>
* feat(composio): per-agent allowlist + originator-scoped MCP overlay (MUL-3869) (#4736)
* feat(composio): per-agent allowlist + originator-scoped MCP overlay (MUL-3869)
Stage 3.1 of the Composio epic (MUL-3721 parent). PR #4704 wired in the
runtime_mcp_overlay column and a per-task dispatch hook; this change
inverts the default from "all-on" to opt-in and locks the overlay to the
agent owner's own connected apps:
- Agents carry composio_toolkit_allowlist TEXT[]. NULL or [] => no MCP.
Owner-only read/write; non-owner GET/PUT silently redacts/drops the
field (same shape as mcp_config).
- agent_task_queue carries originator_user_id UUID. Set from the
top-of-chain HUMAN at every enqueue path:
* issue/mention comment by member -> author_id
* issue/mention comment by agent -> inherit via comment.source_task_id
-> parent task originator_user_id
* quick-create -> requester_id
* chat -> initiator_user_id
* retry -> SQL-inherited from parent row
* autopilot -> NULL (system-driven)
- BuildTaskOverlay (composio dispatch) now takes (ctx, originatorUserID,
agent) and short-circuits on five gates: invalid originator,
originator != agent.owner_id, empty allowlist, empty intersection of
allowlist ∩ active connections, defensive empty session URL. Composio
CreateSession is called with BOTH `toolkits.slugs` (the intersection)
AND `connected_accounts` (the pinned account ids), narrowing the
tool-router twice.
- The originator-vs-owner gate closes the agent-fanout privacy hole: any
workspace member who can @-mention a public agent used to project the
owner's connected apps into their run. Now the overlay only mounts
when the human at the top of the chain IS the agent owner.
Tests:
- dispatch_test.go covers all 5 gates plus uppercase/whitespace slug
normalisation.
- task_runtime_mcp_overlay_test.go covers the no-op gates of the new
applyRuntimeMCPOverlay signature.
- agent_composio_allowlist_test.go (handler): owner roundtrip
(list/empty/null), workspace-admin silent-drop, owner-only GET
visibility, pure normaliseComposioToolkitAllowlist.
- resolve_originator_test.go (service, DB-backed): member-authored,
agent-authored inherits via comment.source_task_id, invalid id.
Migration 129 up/down/up verified against docker postgres.
Co-authored-by: multica-agent <github@multica.ai>
* chore(composio): gofmt + regenerate sqlc with v1.31.1 (MUL-3869 review nits)
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: Eve <eve@multica-ai.local>
Co-authored-by: multica-agent <github@multica.ai>
* fix(composio): accept nested connected account auth config
* feat(views): creator-only MCP tab for per-agent Composio allowlist (MUL-3870) (#4743)
Stage 3.2 frontend on top of the Stage 3.1 backend (MUL-3869,
|
||
|
Bohan Jiang
|
d970b68ce7 |
feat(autopilot): View/Write permission layer + member access delegation (MUL-3807) (#4695)
* feat(autopilot): add View/Write permission layer Autopilot write and execute operations were gated only by workspace membership, so any member could edit, delete, trigger, or rotate the webhook of any autopilot, and GetAutopilot returned webhook tokens to every member (a token alone can trigger the autopilot). - Add canWriteAutopilot / requireAutopilotWrite: update, delete, trigger, replay-delivery, and all trigger/secret management now require the autopilot creator or a workspace owner/admin. - Redact webhook_token/path/url in GetAutopilot for callers without write access; trigger metadata otherwise stays visible (View default = all members). Creating an autopilot stays open to any member. - ANDs with the existing private-assignee-agent dispatch gate. MUL-3807 Co-authored-by: multica-agent <github@multica.ai> * feat(autopilot): delegate write access via collaborators + manage-access UI Adds an explicit grant primitive so an autopilot's creator/admin can authorize specific workspace members to manage it, with a frontend entry point — beyond the implicit creator/owner-admin set from the prior commit. Backend: - New autopilot_collaborator table (migration 128, members-only, app-layer cleanup, no FK) + sqlc queries. - memberCanWriteAutopilot now also honors explicit collaborators; the write gate, webhook-secret redaction, and a new per-caller can_write flag (on list + detail) all flow through it. - POST/DELETE /api/autopilots/{id}/collaborators (writer-gated); GetAutopilot embeds the collaborators list. Delete cleans up grants in its transaction. - Tests: grant->write->revoke flow, non-writer can't grant, non-member rejected. Frontend (web + desktop via packages/views): - ManageAccessDialog: member picker to grant/revoke, current list with remove. - 'Manage access' entry in the autopilot detail header; edit/run/add-trigger/ delete and the list-row kebab + per-trigger rotate/delete now gate on can_write (absent => allowed, server stays the gate). - can_write wired through types/schema/api client/mutations; en + zh-Hans copy. MUL-3807 Co-authored-by: multica-agent <github@multica.ai> * fix(autopilot): add manage-access i18n keys to ja/ko locales The locale parity test requires every non-EN bundle to cover every EN key. The prior commit added detail.manage_access + the access.* block to en and zh-Hans only, failing parity for ja and ko. Add the translated keys to both. Co-authored-by: multica-agent <github@multica.ai> * fix(autopilot): restrict access-list management to creator/admin only Final-review fix: AddAutopilotCollaborator/RemoveAutopilotCollaborator used requireAutopilotWrite, which counts granted collaborators as writers — so a collaborator could in turn grant/revoke others, a privilege escalation contradicting the 'collaborators cannot re-grant' design. - New requireAutopilotAccessManagement guard uses the narrower autopilotWriteByOwnership predicate (creator or workspace owner/admin only); swapped into both collaborator endpoints. Collaborators keep their edit/trigger/secret write-execute rights. - GetAutopilot now also stamps can_manage_access (narrower than can_write); the detail page gates the 'Manage access' button on it so collaborators no longer see an entry that would 403. - Tests: collaborator grant-others -> 403, revoke-peer -> 403, while retaining edit; can_manage_access true for owner, false for collaborator. MUL-3807 Co-authored-by: multica-agent <github@multica.ai> --------- Co-authored-by: J <j@multica.ai> Co-authored-by: multica-agent <github@multica.ai> |
||
|
Multica Eve
|
5d79696fb5 | MUL-3794: rewrite comment routing cascade | ||
|
Bohan Jiang
|
a252f47337 |
fix(scheduler): advance autopilot next_run_at after each scheduled dispatch (MUL-3749) (#4618)
* fix(scheduler): advance autopilot next_run_at after each scheduled dispatch The display-only autopilot_trigger.next_run_at column was written only on trigger create/update and never advanced afterward, so for a recurring schedule it froze at a past slot and the list rendered it as a 'next run' in the past (e.g. '53m ago'). The intended AdvanceTriggerNextRun query was dead code with zero callers. Wire it up at the scheduler's existing post-dispatch seam (replacing the last_fired_at-only TouchAutopilotTriggerFiredAt bump, which AdvanceTrigger- NextRun already supersets). The advanced value is computed on the app local clock via ComputeNextRun — the same path create/update use — so the whole next_run_at display column is owned by one clock and stays consistent; scheduling itself is untouched and still runs off DB time via NextOccurrencesUTC. On a cron/timezone parse failure we fall back to the last_fired_at-only bump. Adds a deterministic regression test for the reported scenario (hourly cron in America/New_York) and documents the local-clock ownership on ComputeNextRun. MUL-3749 Co-authored-by: multica-agent <github@multica.ai> * fix(scheduler): floor next_run_at advance at plan_time to survive clock skew Addresses review feedback on the next_run_at write-back (MUL-3749): - The post-dispatch advance computed the value from time.Now() alone. The handler is entered only after DB time judged the plan due, so if this app instance's clock lags the DB clock at a period boundary, time.Now() could recompute the slot that just fired and next_run_at would not advance — the original staleness bug, at the boundary. Extract advancedNextRun, which anchors at max(now, plan_time) via NextOccurrenceAfterUTC so the written value is always strictly after the fired plan_time while still tracking the local clock in the normal case. - Add scheduler-layer tests asserting the written value is strictly after plan_time across skew / on-slot / normal cases. The previous service-layer test only exercised the helper with an explicit after, not this path. - Sync the stale ListSchedulableAutopilotTriggers comment: the scheduler now writes last_fired_at via AdvanceTriggerNextRun (sqlc regenerated). MUL-3749 Co-authored-by: multica-agent <github@multica.ai> --------- Co-authored-by: J <j@multica.ai> Co-authored-by: multica-agent <github@multica.ai> |
||
|
LinYushen
|
3692b6a862 |
fix(squad): inject leader briefing by task flag, not issue assignee (MUL-3730) (#4606)
* fix(squad): inject leader briefing by task flag, not issue assignee Key squad-leader briefing injection off task.IsLeaderTask + task.SquadID instead of issue.AssigneeType=='squad'. The old gate missed the most common path — an @squad mention in a comment on an issue assigned to a plain agent (MUL-3724) — so the leader booted with zero squad context and did the work itself instead of orchestrating. - migration 127: add agent_task_queue.squad_id (no FK) + partial index - sqlc: CreateAgentTask stamps squad_id; CreateRetryTask inherits it - service: thread squadID through EnqueueTaskForSquadLeader(+WithHandoff), enqueueMentionTask, and the rerun path; all 5 call sites pass the squad id - daemon claim: unified injection keyed on leader-task + squad_id, with a defensive leader-identity re-check; quick-create block retained (it serves issue-less tasks and sets resp.SquadID/SquadName) - briefing: strengthen leader Operating Protocol opening - tests: claim-time injection (comment-mention/non-leader/null-squad), squad_id enqueue stamping, retry inheritance; existing fixture updated Co-authored-by: multica-agent <github@multica.ai> * test+docs(squad): dangling squad_id regression + clarify quick-create path Address review nits on #4606: - Add TestClaim_LeaderTaskWithDanglingSquadID_NoBriefing: squad hard-deleted after enqueue leaves task.squad_id dangling (no FK); claim still 200 and skips injection via the err!=nil guard. This is the load-bearing contract for dropping the FK. - Rewrite the daemon.go injection comment to state quick-create does NOT use the is_leader_task/squad_id columns — it routes squad via the context JSON branch (qc.SquadID) and must not be folded into the column-based path. Co-authored-by: multica-agent <github@multica.ai> --------- Co-authored-by: 魏和尚 <agent@multica.ai> Co-authored-by: multica-agent <github@multica.ai> |
||
|
Multica Eve
|
00b9668cd2 |
fix(autopilot): cold-start planner honors trigger.last_fired_at (MUL-3551) (#4495)
Post-deploy of the new scheduled-dispatch scheduler (PR #4444), an autopilot configured for "weekdays 17:10 Asia/Shanghai" fired at ~12:30 Beijing the day after deploy — ~4h 38m before the next scheduled time the UI showed. Traced to a cold-start regression in the planner hook: Old behaviour ------------- On the first tick after migration the hook found no `sys_cron_executions` row for the trigger (`latestPlan(...).Found == false`) and anchored on the trigger's `created_at`, then applied the 24h replay cap: after := cfg.CreatedAt if oldest := now.Add(-replayWindow); after.Before(oldest) { after = oldest // now - 24h } For a trigger created days/weeks earlier and last fired by the legacy goroutine at Mon 17:10 Beijing (= Mon 09:10 UTC), this set `after = Tue 04:13 UTC - 24h ≈ Mon 04:13 UTC`. The half-open enumeration `(Mon 04:13 UTC, Tue 04:13 UTC]` STILL contained Mon 09:10 UTC — the occurrence the legacy code had already handled — so the new scheduler dispatched it again the moment it took over. The result: a SCHEDULED-source autopilot_run with planned_at = Mon 17:10 Beijing but a wall-clock dispatch at Tue ~12:30 Beijing. Timezone math was correct; the bug was purely the cold-start anchor not respecting prior-fire history. Fix Co-authored-by: multica-agent <github@multica.ai> --- The `autopilot_trigger.last_fired_at` column is maintained by both the legacy goroutine and the new scheduler (via TouchAutopilotTriggerFiredAt), so it is the authoritative "most-recent successful fire" cursor across the migration boundary. The planner hook now anchors cold-start enumeration on it: case latest.Found: after = latest.PlanTime case lastFiredAt != zero: after = lastFiredAt default: after = cfg.CreatedAt For the regressed case, `after = Mon 17:10 Beijing`, the next enumeration window is `(Mon 17:10, Tue 12:30]`, and Tue 17:10 is in the future — the hook returns nothing and the trigger waits quietly for Tue 17:10 as the UI promised. For brand-new triggers (last_fired_at NULL), the original `created_at` path still applies. For long-dormant triggers the `replayWindow` cap remains. Changes ------- * `ListSchedulableAutopilotTriggers` SQL now returns `last_fired_at`. * `autopilotTriggerConfig.LastFiredAt` is populated by the scope provider on every tick. * `autopilotPlansForScope` cold-start branch uses the new anchor. Tests ----- * TestAutopilotScheduleJobColdStartHonorsLastFiredAt — seeds the exact dev-environment shape (created 3 days ago, last_fired_at 5 hours ago, no sys_cron_executions row), runs a tick, asserts zero exec rows AND zero autopilot_run rows. Without the fix this test produces one of each at a historical plan_time. * TestAutopilotScheduleJobColdStartBrandNewTriggerStillFires — asserts a brand-new trigger (last_fired_at NULL) still fires its first due occurrence on cold start. All existing `TestAutopilotScheduleJob*` tests still pass. Refs MUL-3551 Co-authored-by: Eve <eve@multica-ai.local> |
||
|
Multica Eve
|
131ca80a6c |
refactor(autopilot): migrate scheduled dispatch to scheduler.Manager (3/3 MUL-3551) (#4444)
* refactor(autopilot): migrate scheduled dispatch to scheduler.Manager
PR 3 of 3 for the scheduled-Autopilot refactor on MUL-3551.
Replaces the legacy cmd/server/autopilot_scheduler.go goroutine
(30 s app-clock polling, app-time cron advancement, weak crash
recovery) with a JobSpec registered on the existing
scheduler.Manager. sys_cron_executions is now the lease + audit
table for scheduled Autopilot occurrences, and the unique key on
(job_name, scope_kind, scope_id, plan_time) is the primary
guarantee that the same planned fire time cannot produce two runs.
What changed
* server/internal/scheduler/jobs_autopilot.go
New AutopilotScheduleDispatchJob factory:
- scope_kind = "autopilot_trigger", scope_id = trigger.id
- PlansForScope hook (from PR 1) enumerates cron occurrences
in (lastPlan, dbNow] and collapses missed fires to the most
recent one (CatchUpLatestOnly — same policy the legacy
goroutine had, now provable via a one-row-per-tick audit).
- Handler re-loads trigger + autopilot inside the handler so a
between-tick state change (paused, disabled, deleted) takes
effect immediately and is recorded as a no-op SUCCESS row
with skipped_reason in the result JSON.
- Calls AutopilotService.DispatchAutopilotForPlan (from PR 2)
for the actual run creation; that path is itself idempotent
on (trigger_id, planned_at), so a stale-steal retry reuses
the run created by the prior attempt instead of duplicating.
- RunTimeout=2m, StaleTimeout=5m, HeartbeatInterval=30s,
AllowStaleReentry=true, MaxAttempts=3, RetryBackoff
[1m, 5m, 15m], MaxPlansPerTick=5 (safety cap).
* server/internal/scheduler/manager.go
Manager.runOnce promoted to RunOnce (exported) so external test
packages can drive deterministic ticks; existing call sites in
this package + cmd/server tests updated.
* server/internal/service/cron.go
NextOccurrenceAfterUTC and NextOccurrencesUTC: cron evaluators
that take an explicit "now" instant. Callers pass dbNow() so
schedule decisions stay consistent across app instances with
clock skew. Legacy ComputeNextRun is preserved (delegating to
NextOccurrenceAfterUTC with time.Now()) for the display-only
autopilot_trigger.next_run_at write path — scheduling decisions
no longer use it.
* server/pkg/db/queries/autopilot.sql
ListSchedulableAutopilotTriggers replaces the legacy
ClaimDueScheduleTriggers (the new path no longer mutates
autopilot_trigger.next_run_at on claim). RecoverLostTriggers
removed — sys_cron_executions lease theft now handles crash
recovery without an in-handler restart sweep.
* server/cmd/server/main.go
The "go runAutopilotScheduler(...)" line is gone. The new
JobSpec is registered alongside TaskUsageHourlyJob on the
existing schedulerMgr (still using sweepCtx for lifecycle).
* server/cmd/server/autopilot_scheduler.go DELETED.
Tests
* server/internal/service/cron_test.go — unit tests for the cron
helpers: timezone-aware enumeration, half-open (after, until]
window, plan_time-exclusive "after", invalid inputs surface
parse errors, and the "ignores wall clock" property the
scheduler relies on.
* server/cmd/server/autopilot_schedule_job_test.go — DB-backed
integration tests:
- DispatchesOnce: one tick → 1 SUCCESS exec row + 1
autopilot_run with planned_at set; a second tick does not
regress the count.
- MissedSchedulesCollapse: an hour of missed */5 fires
produce a single autopilot_run, not 12.
- CrashRecovery: simulated stale RUNNING lease at the same
plan_time → second tick reclaims it and DOES NOT duplicate
autopilot_run.
- TwoRunnersSingleWinner: two concurrent
scheduler.Manager instances on the same trigger →
per-plan_time uniqueness holds (sys_cron_executions never
has two RUNNING rows at the same plan_time, autopilot_run
count == exec row count).
- DisabledTriggerSkips: a trigger disabled between
scope-list and tick produces no exec row.
- PausedAutopilotSkipsAtHandler: an autopilot paused after
the first tick does not produce a new exec row.
- BadCronFailsLoudly: an invalid cron expression never fires
dispatch (parse error surfaces in the plan hook).
Existing autopilot listener / squad / dispatch tests still
pass.
* server/internal/scheduler/plans_for_scope_test.go from PR 1
still passes (RunOnce rename only).
Verification
* go build ./...
* go vet ./...
* go test ./internal/scheduler ./internal/service ./cmd/server
./internal/handler — all green.
Rollback
* Reverting this commit re-introduces the legacy goroutine.
Migration 124 (PR 2) and the scheduler hook (PR 1) stay in
place. Autopilot data on disk is forward- and backward-
compatible: planned_at columns are nullable, the legacy
goroutine never reads planned_at and the new job never reads
autopilot_trigger.next_run_at.
Refs MUL-3551
Co-authored-by: multica-agent <github@multica.ai>
* fix(autopilot): scheduler hook retries FAILED plans + tighten tests
Review fix for #4444 (MUL-3551).
Blocker: hook planner skipped the FAILED-with-retry plan_time
`autopilotPlansForScope` unconditionally set
`after = latest.PlanTime` when `latest.Found`, then enumerated cron
occurrences in the half-open interval `(after, dbNow]`. That
EXCLUDED the FAILED plan_time itself, so `tryClaim`'s
"FAILED-with-retry" branch — which only fires when the planner
returns the same plan_time — never ran. A claim + crash sequence
left the FAILED row stuck at attempt<max_attempts forever and the
scheduled occurrence was lost (MUL-3551 acceptance ③).
Fix: hook now branches on `latest.RetryEligible(now)` BEFORE
computing `after`. When the most recent stored row is FAILED with
attempts remaining and next_retry_at <= dbNow, the hook returns
`[latest.PlanTime]` unchanged. tryClaim's retry-from-FAILED path
fires, attempt increments, the run is retried, and the audit row
reaches SUCCESS at the same plan_time. Mirrors the cadence
planner's `info.RetryEligible(now)` branch in manager.plansForTick.
Tests
* TestAutopilotScheduleJobCrashRecovery rewritten to actually
pin the retry contract instead of just "no duplicate run":
- assert first attempt completes at attempt=1 with a real
task_id linkage (the "complete" snapshot the retry must
reuse);
- simulate a crash mid-dispatch (status=RUNNING, expired
stale_after, ghost lease_token);
- assert tick 2 transitions the SAME exec row (same plan_time)
to status=SUCCESS at attempt=2 (proving the planner did
NOT skip past the FAILED bucket);
- assert autopilot_run stays at exactly one row, reused from
the first attempt — proving DispatchAutopilotForPlan's
complete-run reuse path is what closes the loop.
* TestAutopilotScheduleJobPausedAutopilotSkipsAtHandler rewritten
to invoke `job.Handler` directly (the previous version drove
`mgr.RunOnce` which short-circuited at the scope-list SQL
filter and never reached the handler). The new test pauses the
autopilot AFTER setup, calls the handler with a fabricated
HandlerInput, and asserts the handler returns
skipped_reason=autopilot_inactive without creating an
autopilot_run.
* TestAutopilotScheduleJobBadCronFailsLoudly renamed to
TestAutopilotScheduleJobBadCronStaysSilent and updated to
match the real implementation: a parse error in the plan hook
surfaces as a manager-level warning log, NOT a
sys_cron_executions row (no plan_time was ever claimed). The
test now asserts zero exec rows AND zero autopilot_run rows,
documenting that bad cron is a permanent configuration error
(caught at HTTP create/update time first), not a transient
failure that belongs in the retry envelope.
Refs MUL-3551
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: Eve <eve@multica-ai.local>
Co-authored-by: multica-agent <github@multica.ai>
|
||
|
Multica Eve
|
eca6102365 |
refactor(autopilot): autopilot_run.planned_at + DispatchAutopilotForPlan (2/3 MUL-3551) (#4443)
* refactor(scheduler): add PlansForScope hook for non-cadence jobs
The current Manager.plansForTick assumes a uniform Cadence grid:
plan_times are derived via FloorPlan(db_now, Cadence). That works for
rollup_task_usage_hourly but not for the upcoming Autopilot schedule
dispatch job, where each trigger has its own cron expression and the
plan_times do not snap to a single global grid.
This change adds an optional JobSpec.PlansForScope hook. When set:
* Manager loads the latest stored plan for (job, scope) and passes
a new LatestPlanInfo to the hook (exported from the previously
private latestPlanInfo). The hook returns the plan_times to attempt
this tick.
* Cadence, CatchUpMode and CatchUpWindow are bypassed; the hook is
in full control of plan_time selection.
* MaxPlansPerTick still acts as a safety cap on the hook's output.
* All other timing fields (RunTimeout / StaleTimeout /
HeartbeatInterval / MaxAttempts / RetryBackoff / AllowStaleReentry)
and the lease/heartbeat/terminal-write SQL primitives are reused
unchanged.
JobSpec.validate now allows Cadence=0 when PlansForScope is set, and
makes the every_plan MaxPlansPerTick > 0 invariant fire only on
Cadence-driven every_plan jobs. Existing rollup_task_usage_hourly
behaviour is unchanged — that JobSpec leaves PlansForScope nil.
Tests:
* TestJobSpecValidatePlansForScopeRelaxesCadence — validate() rules.
* TestManagerPlansForScopeHookDrivesPlans — end-to-end hook delegation
through the manager (DB-backed), proving that hook-returned
plan_times go through the same tryClaim path, MaxPlansPerTick
truncates without erroring, and LatestPlanInfo is populated on the
second tick.
* TestManagerPlansForScopeHookEmptyIsNoOp — empty hook output is a
valid no-op.
No behaviour change for callers that don't set PlansForScope.
Refs MUL-3551
Co-authored-by: multica-agent <github@multica.ai>
* refactor(autopilot): add planned_at + DispatchAutopilotForPlan for occurrence idempotency
PR 2 of 3 for the scheduled-Autopilot refactor on MUL-3551.
Adds dispatch-layer idempotency for scheduled triggers. This is the
second line of defence behind the primary uq_sys_cron_execution
guarantee in sys_cron_executions: if a runner crashes between
"create autopilot_run" and "write SUCCESS in sys_cron_executions",
the next stale-steal retry re-enters dispatch with the SAME
(trigger_id, planned_at). Without a row-level guard, that retry
would create a duplicate autopilot_run, issue, and task.
Changes:
* Migration 124: ALTER TABLE autopilot_run ADD COLUMN planned_at
TIMESTAMPTZ + partial unique index on (trigger_id, planned_at)
WHERE both are NOT NULL. Manual / webhook / api dispatch leaves
planned_at NULL so they keep the existing semantics unchanged.
* autopilot.sql: CreateAutopilotRun now takes planned_at;
GetAutopilotRunByTriggerAndPlanned is the fast-path lookup used
by DispatchAutopilotForPlan to detect a prior attempt's row
without burning an INSERT.
* service.DispatchAutopilotForPlan: new entry point for scheduled
triggers that already know the canonical UTC plan_time of the
occurrence they are firing. Looks up an existing run for
(trigger_id, planned_at) and reuses it on a stale-steal retry;
otherwise dispatches normally with planned_at stamped on the
new run.
* service.DispatchAutopilot keeps its current signature for
manual / webhook / api callers (planned_at stays NULL).
* recordSkippedRun also threads planned_at so the skip path
participates in the same partial-unique guarantee.
* sqlc v1.31.1 regenerated autopilot.sql.go + models.go.
Unrelated workspace.sql.go drift restored.
Tests (against local Postgres):
* TestDispatchAutopilotForPlanIsIdempotent — first call creates a
run; second call with same (trigger, planned_at) reuses it
(autopilot_run row count stays at 1); third call with a different
planned_at on the same trigger creates a second run (proves we
are not collapsing legitimate occurrences).
* TestDispatchAutopilotForPlanRejectsZeroArgs — invalid trigger_id
and zero planned_at both fail loudly so callers cannot silently
disable the idempotency guard.
* Existing autopilot listener / squad / dispatch tests all still
pass.
This PR has no scheduler / handler / UI behaviour change on its own:
the new entry point exists but is not yet wired into the schedule
goroutine. PR 3 will register the autopilot_schedule_dispatch
JobSpec that consumes it and remove the legacy
cmd/server/autopilot_scheduler.go path.
Refs MUL-3551
Co-authored-by: multica-agent <github@multica.ai>
* fix(autopilot): DispatchAutopilotForPlan recovers partial-state runs
Review fix for #4443 (MUL-3551).
Before this change, DispatchAutopilotForPlan returned ANY existing
autopilot_run for (trigger_id, planned_at), including the
half-written rows produced when a runner crashed between
"CreateAutopilotRun" and "create downstream issue/task". The
scheduler handler would then write SUCCESS in sys_cron_executions
even though no issue or agent task was ever created, silently
losing the scheduled occurrence.
Fix:
* New isAutopilotRunComplete helper classifies an existing run:
- terminal status (completed / failed / skipped) → reuse.
- issue_created with valid issue_id → reuse (the issue
listener owns task creation from here).
- running with valid task_id → reuse (the task is queued).
- anything else → partial; must NOT short-circuit.
* New SQL RecoverPartialAutopilotRun marks a partial row FAILED
with a recovery reason AND clears its planned_at. The cleared
planned_at releases the partial-unique slot in
uq_autopilot_run_trigger_planned, letting the fresh dispatch
INSERT a new row at the same (trigger_id, planned_at) without
conflict.
* DispatchAutopilotForPlan now branches on the lookup:
complete run → return; partial run → recover-then-fresh-
dispatch; not-found → fresh dispatch. The fresh dispatch path
still goes through dispatchAutopilot, so the new row carries
the real issue_id / task_id by the time the handler returns.
* Tests: TestDispatchAutopilotForPlanRecoversPartialRun seeds a
partial run (status='running', task_id=NULL for run_only;
status='issue_created', issue_id=NULL for create_issue) and
asserts the retry:
- returns a DIFFERENT run row (no false reuse);
- leaves the partial row in status='failed', planned_at=NULL,
with a non-empty failure_reason for ops;
- produces a fresh row with planned_at preserved AND the
appropriate downstream linkage (task_id for run_only,
issue_id for create_issue);
- exactly one live row at (trigger_id, planned_at) after
recovery, so the partial-unique constraint is honoured.
Existing TestDispatchAutopilotForPlanIsIdempotent and
TestDispatchAutopilotForPlanRejectsZeroArgs still pass — the
complete-reuse path is unchanged for the realistic SUCCESS-state
case.
Refs MUL-3551
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: Eve <eve@multica-ai.local>
Co-authored-by: multica-agent <github@multica.ai>
|
||
|
Bohan Jiang
|
5038c983c0 |
MUL-3281: Add daemon skill bundle refs (#4445)
* feat: add daemon skill bundle refs Co-authored-by: multica-agent <github@multica.ai> * fix: tighten skill bundle resolve safeguards Co-authored-by: multica-agent <github@multica.ai> * feat: add task prepare lease Co-authored-by: multica-agent <github@multica.ai> * fix: isolate prepare lease concurrent index migration Co-authored-by: multica-agent <github@multica.ai> * fix: keep prepare lease active through start Co-authored-by: multica-agent <github@multica.ai> --------- Co-authored-by: J <j@multica.ai> Co-authored-by: multica-agent <github@multica.ai> |
||
|
Naiyuan Qing
|
4ab335b8a5 |
MUL-3416: Issue pre-trigger preview + Handoff Note (#4383)
* feat(issues): unify run-enqueue decision behind WillEnqueueRun + preview endpoint Collapse the issue update/batch enqueue copies into one service predicate service.IssueService.WillEnqueueRun, shared verbatim with a new dry-run endpoint POST /api/issues/preview-trigger so the four entry points stop drifting (squad/self-loop/batch omissions, MUL-3375). The private-agent gate stays at the HTTP boundary: write paths inject allow-all, preview injects the real gate so it never leaks a private agent's readiness. Add suppress_run to issue update/batch: the change applies but no run starts. Remove the now-dead handler mirrors shouldEnqueueSquadLeaderOnAssign / isSquadLeaderReady. service.Create and the comment trigger chain are untouched. Tests: preview behavior, preview<->write-path match, batch aggregation, member no-trigger, suppress_run skip, malformed-body 400. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Co-authored-by: multica-agent <github@multica.ai> * feat(issues): inject handoff note into assigned runs via first-class task field Add an optional handoff_note carried by issue assign/promote into the run's opening prompt and issue_context.md, via a dedicated agent_task_queue column (migration 122) and a daemon assignment-handoff render branch — never a fabricated comment, never trigger_comment_id (MUL-3375 §6.1). Thread the note through enqueueIssueTask/enqueueMentionTask + WithHandoff public variants and dispatchIssueRun; suppress_run or a parked write drops it (no run = nothing to inject). Soft version gate: MinHandoffCLIVersion + HandoffSupported, surfaced per-trigger as handoff_supported in the preview so the UI can gray the note box on old daemons; the assignment never hard-fails. Tests: daemon prompt + issue_context render via the assignment branch (not quick-create/comment), version helper matrix, note persists on the task, suppressed assign enqueues nothing. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Co-authored-by: multica-agent <github@multica.ai> * feat(issues): leave a display-only handoff record on the timeline When an assign/promote with a handoff note starts a run, write one type='handoff' timeline record via TaskService.RecordHandoff — a direct Queries.CreateComment + timeline event that bypasses Handler.CreateComment, so it never reaches triggerTasksForComment and cannot start a second run (MUL-3375 §6.2, the must-not-retrigger invariant). Author is the actor who handed off; body is the note. Migration 123 admits the 'handoff' comment type. Recorded only on a real run start: suppress_run or a parked write writes nothing. enqueueSquadLeaderTask now reports whether it enqueued so the trace is gated on an actual dispatch. Test: exactly one handoff record on assign-with-note, exactly one task (no re-trigger), and no record when suppressed. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Co-authored-by: multica-agent <github@multica.ai> * feat(issues): frontend plumbing for issue-trigger preview + handoff (core) Add api.previewIssueTrigger + IssueTriggerPreviewSchema (zod parseWithFallback), the use-issue-trigger-preview hook, issueKeys.issueTriggerPreview(+All) with WS queue-state invalidation, suppress_run/handoff_note on UpdateIssueRequest, the 'handoff' CommentType, and stripping of the control fields from optimistic update/batch cache patches (MUL-3375 §9). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Co-authored-by: multica-agent <github@multica.ai> * fix(issues): exclude handoff records from new-comment counting type='handoff' is a display-only timeline record, not conversation. Exclude it from CountNewCommentsSince so a handoff note never inflates the count of "new comments to catch up on" fed to a claiming agent (MUL-3375 §12). Analytics already excludes it (RecordHandoff is a direct write that emits no analytics event), and the comment-trigger path is already bypassed. Test: a handoff record does not bump the new-comment count; a real comment does. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Co-authored-by: multica-agent <github@multica.ai> * feat(issues): pre-trigger preview UI, handoff note, timeline card (web/desktop) Wire the §9 frontend onto the preview endpoint + handoff fields: - Delete the backlog blocking dialog (backlog-agent-hint*) and its modal type; the over-eager nag is gone. Backlog awareness is now a passive label. - RunConfirmModal: single assign + batch assign/status route here. Shows the backend predicate's verdict ("将启动 @X" / "将启动 N 个" / parked), an optional handoff note (assign only, soft-gated by handoff_supported), and 暂不启动 — then applies via update/batch. No frontend guessing. - create modal: passive CreateRunHint ("将启动 @X" / backlog parked). - single status change stays a direct apply (unchanged). - timeline: render type='handoff' as a distinct, non-interactive handoff card. - i18n run_confirm + handoff_card across en/ja/ko/zh-Hans; drop backlog action keys; locale parity green. Tests: use-issue-actions (assign → run-confirm modal, member → direct), create-issue + comment-card suites updated/green; views typecheck + lint clean. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Co-authored-by: multica-agent <github@multica.ai> * test(issues): use a valid anchor in the handoff count-exclusion test CountNewCommentsSince filters id <> @anchor_id; SQL id <> NULL is NULL and excludes every row, so an empty anchor made the control assertion read 0. The production caller always passes a real anchor — mirror that with a non-matching sentinel uuid. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Co-authored-by: multica-agent <github@multica.ai> * test(issues): RunConfirmModal apply logic (start/suppress/note-gate/batch) Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Co-authored-by: multica-agent <github@multica.ai> * test(core): preview schema malformed/missing/null fallback coverage Cover IssueTriggerPreviewSchema via parseWithFallback (MUL-3375): well-formed parse, top-level + item default fills (empty/older backend), and fallback to { triggers: [], total_count: 0 } for malformed shapes, a dropped required issue_id, a wrong-typed total_count, and null/non-object bodies — so the four entry points degrade to "nothing will start" instead of throwing. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Co-authored-by: multica-agent <github@multica.ai> * refactor(issues): remove display-only handoff timeline record (留痕) The handoff "留痕" timeline record (type='handoff' comment written on run start) was judged superfluous and dropped per product call. This removes only the display-only trace; the handoff NOTE injection into the run's opening prompt + issue_context.md is untouched. - backend: drop RecordHandoff + its call in dispatchIssueRun - db: drop the `type <> 'handoff'` exclusion in CountNewCommentsSince and migration 123 (comment_type_check reverts to the 4-type set from 001); no production data exists for this unreleased feature - frontend: drop the "handoff" CommentType, HandoffCard, and handoff_card i18n (all locales) - tests: drop handoff_count_test.go and the record-write assertions in issue_trigger_preview_test.go (note-injection tests retained) Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Co-authored-by: multica-agent <github@multica.ai> * feat(issues): dismissable run-confirm modal + team-handoff copy Two fixes to the pre-trigger confirm modal (MUL-3375). 1. Dismissable: switch RunConfirmModal from AlertDialog to the standard shadcn Dialog so it has the close (X) button + Esc + click-outside. Previously the only choices were "start" / "don't start now" with no way to abort the action entirely; dismissing now cancels with no write. 2. Copy: rework the action-surface wording away from the backend term "run" toward team-handoff voice — 指派 / 开始 / 交接 (run stays only on record surfaces). Unifies the note's three names to "交接说明", and parallels the rewrite across en/ja/ko. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Co-authored-by: multica-agent <github@multica.ai> * chore(agent): bump handoff note min CLI version to 0.3.28 The daemon release that renders handoff notes ships in 0.3.28 (0.3.27 was the prior tag), so move the soft-gate threshold up. Below this the note is silently dropped and the frontend grays the note box — assignment is never blocked. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(issues): skip run-confirm when batch-moving issues to backlog A move into backlog never starts a run (service/issue_trigger.go), so the pre-trigger confirm modal degenerated to an empty "won't start" box with a single Apply button — pure friction. Apply directly instead, matching the single-issue status path. Other target statuses still route through the modal. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * feat(issues): refine pre-trigger preview hint and copy - Move the create-issue run hint to a reveal band (grid 0fr→1fr) above the property toolbar. It was sharing the footer button row and, lacking a width constraint, reflowed the submit buttons whenever it appeared. Restyle to a borderless, comment-style avatar+caption that is purely a caption (non-interactive avatar). - Distinguish squad from agent in the pre-trigger copy: a squad's leader evaluates and delegates rather than "starting work" itself. Add will_start_named_squad / will_start_squad / create_will_start_squad across en/zh/ja/ko (reusing the squad_leader_* evaluate→arrange vocabulary) and branch run-confirm + the create hint on squad assignees. - Bold the assignee name in the run-confirm headline via a language-safe sentinel split (no per-language prefix/suffix keys). - Align zh "开始处理" → "开始工作" on the single-assign copy. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * test(issues): stub ActorAvatar in create-issue suite CreateRunHint now renders an ActorAvatar for agent/squad assignees, which pulls in getActorInitials/getActorAvatarUrl + the workspace/presence/navigation hook tree. This form-focused suite only stubbed getActorName, so the squad-forwarding test crashed with "getActorInitials is not a function". Stub the avatar inert — its own behavior is covered elsewhere. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Walt <walt@multica.ai> Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com> Co-authored-by: multica-agent <github@multica.ai> |
||
|
Jiayuan Zhang
|
2d71872daa |
feat(autopilot): default subscriber template (MUL-2533) (#3060)
* feat(autopilot): default subscriber template (MUL-2533) — server Add per-autopilot member subscriber template that fans out to every issue the autopilot spawns. New autopilot_subscriber table; extend issue_subscriber.reason with 'autopilot' so the dispatch-time fanout is distinguishable from manual subscriptions. API: POST/PATCH /api/autopilots accept a `subscribers` array (member user_type only for the first version); PATCH semantics are full-replace. GET returns subscribers on the detail endpoint; the list endpoint omits them to avoid an N+1. Dispatch: dispatchCreateIssue lists the template inside the same tx as the issue insert and writes the rows with reason='autopilot' before EventIssueCreated fires, so notification listeners see the full subscriber set on the first event. Co-authored-by: multica-agent <github@multica.ai> * feat(autopilot): default subscriber template (MUL-2533) — frontend New SubscriberMultiSelect picker (members-only search + chips) wired into the create / edit AutopilotDialog. The detail page renders the saved template as read-only chips; edits flow through the dialog. TS types expose the new `subscribers` field on Autopilot, plus an AutopilotSubscriberInput shape for the create/update wire payloads. Co-authored-by: multica-agent <github@multica.ai> * fix(autopilot): notify template subscribers on issue creation (MUL-2533) The autopilot create-issue path fans out template subscribers into issue_subscriber inside the same tx as the issue insert, but the issue:created notification listener only matches handler.IssueResponse payloads and only direct-notifies the assignee + @mentions. The autopilot publishes a map[string]any payload, so the listener falls through and the template subscribers never receive an inbox item for the creation event — breaking OQ3 ("reason='autopilot' subscribers receive all subscription events, consistent with reason='manual'"). Fix it where the divergence lives: in dispatchCreateIssue, right after EventIssueCreated fires, write an inbox_item (type='issue_subscribed', severity='info') for each member subscriber and publish EventInboxNew so the recipient's inbox WS feed updates in real time. The write is after the tx commit so an inbox hiccup can't roll back the issue; failures are logged, not propagated. The manual path is unchanged — manual subscribers don't exist at creation time, so there is nothing to notify there. Adds a new InboxItemType 'issue_subscribed' (en/zh labels) and two covering tests in autopilot_subscriber_test.go: one asserts the inbox row lands for a template subscriber on dispatch, the other asserts the no-subscriber autopilot stays silent. Co-authored-by: multica-agent <github@multica.ai> * fix(autopilot): align subscriber PR with current main Co-authored-by: multica-agent <github@multica.ai> * fix autopilot subscriber template transaction Co-authored-by: multica-agent <github@multica.ai> --------- Co-authored-by: Lambda <lambda@multica.ai> Co-authored-by: multica-agent <github@multica.ai> |
||
|
Naiyuan Qing
|
63cf0ed308 |
feat(lists): rebuild all six list surfaces on a shared Linear-style list grid (#4038)
* fix(issues): render thread replies in chronological order (#3691)
collectThreadReplies walked the parent_id tree depth-first, so an agent
reply forced to nest under its trigger comment rendered before earlier
sibling replies (A-D-B-C instead of A-B-C-D) whenever the agent returned
late. Sort the collected subtree by created_at (id tie-break) so the
thread reads in arrival order — the same order the server already feeds
agents via `comment list --thread` (ListThreadCommentsForIssue).
All other consumers of the array (resolution derivation, fold bars,
counts, deep-link) are order-independent.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* feat(skills): rebuild skills list on shared Linear-style list grid
- new ListGrid primitives (subgrid: single source of truth for column tracks)
- skills list: sortable columns, used-by avatar stack, source/creator columns,
row kebab + batch toolbar with add-to-agent and delete
- skill view store in core; addAgentSkills client method; HoverCheck extracted
to views/common (issues header now imports the shared copy)
- locale keys for list actions/filters and the reworked detail page
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* feat(skills): rework detail page into overview/files tabs
- tabs directly under the breadcrumb header: overview (default) and files
- overview: identity block + rendered SKILL.md as the main column, right
rail with metadata card (source/creator/updated, inline name+description
edit toggle) and used-by panel with bind/unbind
- files: file tree + viewer/editor unchanged; SKILL.md "edit" jumps here
- header kebab menu (copy skill ID, delete); page-level save bar shared by
both tabs; tab state persisted in ?tab=
- file tree: ARIA tree roles + roving-tabindex keyboard navigation
- drop the old right sidebar (metadata dl, permissions paragraph)
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* revert(skills): restore detail page to main, keep branch list-only
Drop the overview/files tabs rework from this branch so the PR scope is
the list rebuild only. skill-detail-page.tsx and file-tree.tsx are back
to the main versions; the locale detail/file_tree sections are restored
to match. The detail rework is preserved on stash/skills-detail-tabs
for a follow-up PR.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* feat(skills): drop description column from skills list
Description is agent-facing routing metadata, not a scannable list
property — Linear's display options expose no description column for
the same reason. Removes the cell, column key, display toggle, lg grid
track, skeleton cells, and the now-dead table.description /
table.no_description locale keys.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* feat(skills): drive list column hiding by container width, drop by priority
Replace viewport sm:/lg: breakpoints with Tailwind v4 container query
variants (@2xl/@4xl) on the list wrapper, so an open sidebar or split
pane narrows the column set instead of squashing tracks. Remove the
min-w-fit + overflow-x-auto horizontal-scroll fallback: when space runs
out, low-priority columns (created/source/creator, then updated) drop
and return as the container widens; name and usedBy never drop. ListGrid
conventions comment updated — this is the template for all list pages.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* feat(skills): virtualize list rows with @tanstack/react-virtual
Linear-style headless virtualization: the virtualizer computes the
visible index range and offsets; offsets land as padding on the
scrolling ListGridBody so mounted rows stay direct subgrid children and
column alignment is untouched. Fixed 48px rows skip per-row measurement.
Hideable column tracks move from max-content to deterministic widths
(CSS vars) — with only the visible slice mounted, content-driven tracks
would resize during scroll. A user-hidden column zeroes its var so the
track still collapses; per-cell max-w caps move into the tracks.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* fix(skills): list tiers must fit their container trigger width
The @4xl tier's track sum (~1080px with gaps) exceeded its 896px
trigger; with the horizontal-scroll fallback gone, the right-side
columns were clipped unreachably between 896-1080px. Move tier 3 to
@5xl (1024px), trim usedBy/source/creator tracks, and document the
fit invariant with its arithmetic next to the template and in the
ListGrid conventions.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* feat(skills): show description as subtext under the skill name
Lives in the name track as a second truncated line (max-w 36rem,
title attr for the full text) — no track, no header, no slot in the
responsive arithmetic. Both lines fit the fixed 48px row, so the
virtualizer contract is untouched; rows without a description center
the name.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* Revert "feat(skills): show description as subtext under the skill name"
This reverts commit
|
||
|
Bohan Jiang
|
24b162cdbc |
feat(daemon): surface the real task initiator to the agent runtime (MUL-2645) (#3899)
* feat(daemon): surface the real task initiator to the agent runtime (MUL-2645)
In a multi-person workspace the agent runtime only ever saw the runtime
OWNER identity: the brief's `## Requesting User` is sourced from
runtime.OwnerID and the task-scoped token is owner-bound, so every
requester (whoever commented, @mentioned, or chatted) appeared to the
agent as the owner. Agents that route by initiator for permission,
privacy, or audit all misjudged.
Resolve the real task initiator at claim time and surface it distinctly
from the owner:
- comment / mention trigger -> triggering comment's author (member or agent)
- chat task -> chat session creator (sessions are creator-only)
- on-assign / autopilot / quick-create -> no attributable initiator (omitted)
Adds initiator_{type,id,name,email} to the claim response, the daemon
Task, and TaskContextForEnv, rendered into the brief as a new
`## Task Initiator` section. The section documents the privacy boundary:
the agent's credentials stay owner-scoped, so this is an attested
identity for the agent's own routing/privacy logic, not act-as. No DB
migration — both paths are derivable from existing rows.
Tests: brief rendering (member/agent/omit/sanitize) + email guard unit
tests, and claim-handler tests for the comment and chat paths.
Co-authored-by: multica-agent <github@multica.ai>
* fix(chat): store real sender as task initiator, not chat_session creator (MUL-2645)
Review fix (Niko, PR #3899). v1 resolved the chat task initiator from
chat_session.creator_id at claim time. That is correct for web chat and
Lark p2p (creator == sender), but WRONG for Lark group chats: the group
session creator is deliberately the installer (stable identity across
member churn), not the message sender. So in a Lark group, every member
who triggered the agent showed up in the brief as the installer/owner —
the exact bug this issue is about, still live at that entry point.
Capture the real sender at enqueue time instead of deriving it from the
session creator at claim time:
- migration 117: agent_task_queue.initiator_user_id (FK user, ON DELETE
SET NULL); NULL for non-chat and pre-migration rows.
- EnqueueChatTask now takes an explicit initiatorUserID. Web chat passes
the authenticated request user; the Lark dispatcher threads the inbound
sender (binding.MulticaUserID) through scheduleRun -> flushChatRun. The
debouncer keeps the latest scheduled flush per session, so in a multi-
sender silence window the LATEST sender wins (documented + tested).
- claim handler resolves the initiator from task.initiator_user_id and
drops the creator_id fallback entirely.
The Lark group session creator stays the installer (unchanged) — only the
task initiator is corrected, keeping the two concepts cleanly separate.
Tests: dispatcher group regression (initiator = sender, not installer),
latest-sender-wins, p2p initiator assertion; the chat claim handler test
now sets creator != initiator and asserts the stored sender wins.
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: J <j@multica.ai>
Co-authored-by: multica-agent <github@multica.ai>
|
||
|
Raúl Anatol
|
2b5696703f |
MUL-2703: feat(autopilots): webhook event filters per trigger (MUL-2334 follow-up) (#3231)
* feat(autopilots): webhook event filters per trigger (MUL-2334 follow-up) Adds schema-backed event/action filtering to webhook triggers so operators can declare exactly which GitHub (or generic) events should spawn autopilot runs. Events outside the declared scope are recorded as ignored with reason 'event_filtered' — visible in the delivery log but without expensive run/task creation. Closes #3093 (supersedes the description-parsing approach from that PR). Backend: - Migration 108 adds event_filters JSONB to autopilot_trigger - sqlc queries updated for CREATE / UPDATE / LIST / GET - HandleAutopilotWebhook filters against trigger.event_filters before dispatch - Create/Update trigger handlers accept event_filters in the request body - Response shape includes event_filters so the UI can render it Frontend: - New WebhookEventFilterSection component in the autopilot dialog - Inputs for event name + comma-separated actions - i18n strings added (en + zh-Hans) Tests: - Unit tests for splitWebhookEvent and webhookEventAllowedByTriggerScope - Handler-level integration tests for filtered / allowed / no-filter paths co-authored-by: ZephaniaCN <agent/autopilot-webhook-filter> * fix: recognize gitlab/bitbucket/gitea as providers in splitWebhookEvent TestSplitWebhookEvent failed because only 'github' was recognized as a provider prefix. Extract isKnownProvider() to handle gitlab, bitbucket, and gitea as well. * fix(autopilots): address PR #3231 review for webhook event filters Must-fix from PR #3231 review: 1. event_filters now uses typed []WebhookEventFilter at the HTTP boundary instead of []byte. encoding/json was base64-encoding the field on the way out, so the UI could not .map() the response, and a real JSON array on the way in failed to decode. Response field also decodes the stored JSONB into a typed slice before serialising back. 2. UpdateAutopilotTriggerRequest.EventFilters is *[]WebhookEventFilter with tri-state PATCH semantics: nil pointer = leave alone, [] = clear, [...] = replace. The handler marshals an explicit empty slice to the JSONB literal `[]` so COALESCE overwrites instead of preserves. AutopilotDialog now PATCHes the webhook trigger when event_filters change in edit mode (previously the toast said "updated" while the backend was unchanged). 3. webhookEventAllowedByTriggerScope no longer short-circuits to false on the first event-name match whose actions don't line up. Earlier code silently shadowed any later filter that shared the same event name with disjoint actions. Robustness: validateWebhookEventFilters rejects empty event names / actions at write time, and the matcher fails closed on malformed stored bytes instead of widening the allowlist. Tests: handler tests now post real JSON arrays (the prior []byte path masked the contract bug). Adds round-trip / clear-with-[] / preserve- when-omitted / replace / invalid-filter / filters-on-schedule coverage, plus matcher tests for same-event multi-filter and malformed-deny. Migration renamed 108 → 110 to avoid colliding with main's 108_task_token (came in via the merge from main). |
||
|
Bohan Jiang
|
341ce7bfa5 |
feat: support local working directory for projects (MUL-2618 v1) (#3283)
* feat(project): add local_directory project_resource type (MUL-2662)
Adds a second project_resource type alongside github_repo so a project
can be pinned to an existing directory on a specific daemon (the v1 of
the local-working-directory flow tracked in MUL-2618). The ref schema is
{ local_path, daemon_id, label? }; local_path must be absolute and
daemon_id is required. The same (daemon_id, local_path) pair is allowed
on multiple projects by design — no UNIQUE constraint is added.
Implementation reuses the existing project_resource API surface: the new
type is wired through the validator switch with no migration, no new
events, and no daemon-handler changes (daemon already passes through
arbitrary resource types via ProjectResources). The CLI gains
--local-path / --daemon-id / --ref-label shortcuts so
`multica project resource add --type local_directory` mirrors the
existing `--type github_repo --url ...` ergonomics; the generic --ref
flag still works for both types.
Tests cover the full CRUD lifecycle, the same-path-across-projects
allowance, the same-path-same-project conflict, the validator rejections
(missing/blank/relative path, missing daemon_id, wrong payload type),
and the cross-platform isAbsoluteLocalPath helper.
Co-authored-by: multica-agent <github@multica.ai>
* feat(project): add update endpoint + label-shadow guard for project_resource (MUL-2662)
Addresses the Elon review on PR #3263:
- Add PUT /api/projects/{id}/resources/{resourceId} with sqlc query,
matching handler, CLI `project resource update`, and a new
EventProjectResourceUpdated WS event. resource_type stays immutable;
ref/label/position are all individually optional.
- Catch same-project (daemon_id, local_path) collisions where only the
embedded label differs — the row-level UNIQUE only matches the full
ref JSON, so a label typo would otherwise let the same working
directory bind twice.
- Tests cover the update lifecycle (label-only / ref / clear / 404 /
invalid path) and the label-shadow conflict on both create and
update; the in-place rename still succeeds because the conflict
scan ignores the row being edited.
Incidental: regenerating sqlc picked up a missing skills_local scan in
UpdateAgentCustomEnv that drifted in from #3200.
Co-authored-by: multica-agent <github@multica.ai>
* fix(project): close bundled-create label-shadow gap + merge resource_ref on CLI update (MUL-2662)
Two follow-ups from MUL-2662 review round 2:
- CreateProject inline resources path now dedupes local_directory entries on
(daemon_id, local_path) before opening the transaction. The DB-level
UNIQUE(project_id, resource_type, resource_ref) constraint only fires on a
full JSON match, so two rows with the same target but different `label`
would otherwise slip past. Standalone POST/PUT already cover this via
findLocalDirectoryConflict; bundled create was the missing surface.
- `multica project resource update` now seeds resource_ref from the existing
row before applying per-type shortcut flags, so `--default-branch-hint x`
on its own no longer constructs a payload missing `url` (which the server
400s on). Local_directory partial edits get the same merge behavior.
Co-authored-by: multica-agent <github@multica.ai>
* feat(desktop): local_directory project_resource UI (MUL-2665) (#3273)
* feat(desktop): local_directory project_resource UI (MUL-2665)
First UI surface for the local-working-directory flow tracked in MUL-2618.
Lets users on the desktop pin a project to an existing folder on this
machine; web stays read-only since the per-daemon check can't be done in
the browser.
What's new for the renderer:
- ProjectResourcesSection grows a desktop-only "Add local directory"
button next to the existing GitHub-repo popover. Clicking it opens
Electron's native folder picker, validates the path through a new
IPC pair (existence + r/w), and submits a project_resource of
resource_type=local_directory with daemon_id pulled live from
daemonAPI.getStatus.
- LocalDirectoryRow renders the rename pencil + path tooltip, and
greys out when ref.daemon_id != this machine's daemon_id (with a
"only available on the machine that registered this directory"
tooltip). Delete stays enabled so users can drop stale registrations
from any device.
- LocalDirectoryHint sits above the issue-detail comment composer and
shows "Agent will work in-place at {label} ({path})" when the issue's
project has a local_directory matching this daemon. Hidden on web.
- TaskStatusPill picks up a new "waiting_for_directory_release" stage
that the daemon will publish when it dequeues a task but can't
acquire the path lock. The render is in place now so the daemon
sibling subtask can wire the status string without an additional UI
PR.
Plumbing:
- @multica/core/types gains LocalDirectoryResourceRef +
UpdateProjectResourceRequest, and the api client gets the matching
PUT method backed by the server endpoint that landed in
|
||
|
LinYushen
|
5bacfd9742 |
MUL-2526 feat: add member(user_id, workspace_id) index + upgrade sqlc to v1.31.1 (#3046)
- Add migration 106: CREATE INDEX CONCURRENTLY on member(user_id, workspace_id) - Rewrite ListWorkspaces to drive from member table with explicit fields - Regenerate all sqlc code with v1.31.1 (intentional version upgrade) Co-authored-by: multica-agent <github@multica.ai> |
||
|
Angular
|
1f978bf1ec |
feat(autopilot): link created issues to projects (#2908)
* feat(autopilot): link created issues to projects * test(autopilot): cover project flag |
||
|
Jiayuan Zhang
|
fc8528d64d |
feat(autopilot): support assigning to a squad (MUL-2429) (#2888)
* feat(autopilot): support assigning autopilot to a squad (MUL-2429) Path A (Squad-as-Leader) from the RFC: when an autopilot's assignee is a squad, dispatch resolves to squad.leader_id and executes against the leader's runtime — semantics match a human manually assigning the issue to that squad, no fan-out. Backend scope only; frontend picker change is a follow-up PR. Changes: - 096_autopilot_squad_assignee migration: drop agent FK on autopilot.assignee_id, add assignee_type column (default 'agent'), add autopilot_run.squad_id attribution column. - service.AgentReadiness: single source of truth for archived / runtime-bound / runtime-online checks. Shared by autopilot admission gate, run_only dispatch, and isSquadLeaderReady. - service.resolveAutopilotLeader: translates assignee_type/id to the agent that actually runs the work. - dispatchCreateIssue: stamps issue with assignee_type='squad' for squad autopilots and enqueues via EnqueueTaskForSquadLeader. - dispatchRunOnly: belt-and-braces readiness re-check after resolving squad → leader so a leader that went offline between admission and dispatch produces a clean failure instead of a doomed task. - handler.CreateAutopilot / UpdateAutopilot: accept assignee_type with squad/agent existence + leader-archived validation. Backward-compatible default of "agent" preserves the contract for older clients. - Analytics: AutopilotRunStarted/Completed/Failed events carry assignee_type and squad_id; PostHog can now group autopilot runs by squad without joining back to the autopilot row. Co-authored-by: multica-agent <github@multica.ai> * fix(autopilot): reject archived squads, route post-admission skips, cleanup dangling-agent autopilots (MUL-2429) Addresses three review findings on PR #2888: 1. Archived squad handling: validateAutopilotAssignee now rejects squads with archived_at set; resolveAutopilotLeader returns errSquadArchived so the admission gate fails closed; DeleteSquad now mirrors the issue transfer for autopilot rows (TransferSquadAutopilotsToLeader) so surviving autopilots flip to assignee_type='agent' (leader) instead of dangling at the archived squad. 2. dispatchRunOnly post-admission readiness: introduces errDispatchSkipped sentinel, recognised by DispatchAutopilot via handleDispatchSkip so the run is recorded as `skipped` (not `failed`). Manual triggers no longer 500 when the leader's runtime goes offline between admission and task creation. New TestManualTriggerDoesNotErrorOnPostAdmissionSkip locks the behaviour in. 3. Dangling agent assignee after migration 096 dropped the FK: shouldSkipDispatch now distinguishes pgx.ErrNoRows / errSquadArchived (hard skip — retrying won't help) from transient DB errors (fail-open). DeleteAgentRuntime pauses autopilots that target agents about to be hard-deleted (ListArchivedAgentIDsByRuntime + PauseAutopilotsByAgentAssignees) so the breakage surfaces as a paused row in the UI instead of a quiet skip-burning loop. Unit tests cover the sentinel unwrap contract and errSquadArchived errors.Is behaviour. Integration test TestAutopilotDispatchSkipsWhenRuntimeOffline re-verified against a fresh DB with migration 096 applied. Co-authored-by: multica-agent <github@multica.ai> * fix(autopilot): bump last_run_at on post-admission skip (MUL-2429) Match recordSkippedRun (pre-flight skip) and the success path so the scheduler / "last seen" UI both reflect that this tick evaluated the trigger, even when the post-admission readiness gate caught a late regression. Addresses Emacs review caveat #1 on PR #2888. Co-authored-by: multica-agent <github@multica.ai> * feat(autopilot): mixed agent/squad assignee picker in dialog (MUL-2429) End-to-end UI for assigning an autopilot to a squad. Closes the PR #2888 backend gap: the squad-as-assignee feature was already wired in Go (Path A, RFC §4) but the desktop dialog never offered the choice. - core/types/autopilot: add `AutopilotAssigneeType`, surface `assignee_type` on `Autopilot` + Create/Update request payloads. - views/autopilots/pickers/agent-picker: switch to a polymorphic AssigneeSelection (`{type, id}`); render agents and squads as two grouped sections with shared pinyin search. - views/autopilots/autopilot-dialog: maintain `assigneeType` state, send it on create/update, render the trigger avatar / hover dot with `assignee.type`. - views/autopilots/autopilots-page + autopilot-detail-page: render the assignee row using `autopilot.assignee_type` so squad-typed autopilots show the squad avatar + name, not a broken agent lookup. - locales: add `agents_group` / `squads_group` / `select_assignee` keys (en + zh-Hans), keep legacy `select_agent` for callers that still reference it. Co-authored-by: multica-agent <github@multica.ai> --------- Co-authored-by: Lambda <lambda@multica.ai> Co-authored-by: multica-agent <github@multica.ai> |
||
|
Bohan Jiang
|
2323b72710 |
feat(autopilots): webhook delivery layer + idempotency/signature/replay (MUL-2334) [PR1] (#2774)
* feat(autopilots): webhook delivery layer + idempotency / signature / replay (MUL-2334)
Splits "inbound webhook receipt" from "autopilot run creation" so we can
record duplicate attempts, signature outcomes, and ignored/skipped
deliveries — and replay a delivery on demand. v1 ingress wrote straight
into autopilot_run.trigger_payload, which collapsed the two concerns and
left run_only autopilots vulnerable to provider retry storms.
Backend only (PR1). UI Deliveries tab follows in PR2.
Schema (migration 093):
- autopilot_trigger.provider: 'generic' | 'github' (default 'generic').
- autopilot_trigger.signing_secret: nullable plaintext (HMAC needs it
cleartext; mirrors how webhook_token is stored).
- webhook_delivery: one row per inbound POST. Carries raw_body,
selected_headers, dedupe_key/source, signature_status,
autopilot_run_id, replayed_from_delivery_id, response_status / body.
- Partial unique index on (trigger_id, dedupe_key) excludes NULL and
'rejected' rows, so a wrong-secret 401 does NOT permanently block a
future retry with the same X-GitHub-Delivery once the operator fixes
the secret.
Ingress flow (autopilot_webhook.go), persist-first + sync dispatch:
1. IP rate limit -> 2. token lookup -> 3. token rate limit ->
4. read raw body -> 5. autopilot/workspace cross-check ->
6. normalize JSON (400 without persistence on parse failure) ->
7. compute dedupe key + signature status ->
8. INSERT delivery (status=queued). On (trigger_id, dedupe_key)
unique-violation: bump attempt_count on existing row and return
the original delivery_id + autopilot_run_id with 200 ->
9. invalid/missing signature: UPDATE -> rejected, return 401 with
delivery_id (no dispatch, not replayable) ->
10. trigger disabled / autopilot paused/archived: UPDATE -> ignored,
return 200 ->
11. DispatchAutopilot synchronously, UPDATE -> dispatched/skipped/failed
with autopilot_run_id and the response body we returned ->
12. TouchAutopilotTriggerFiredAt and return 200.
No new long-running worker. A stale 'queued' row only happens if the
process dies between INSERT and UPDATE; that's a follow-up sweeper, not
this PR.
Authenticated API:
- GET /api/autopilots/{id}/deliveries (slim list)
- GET /api/autopilots/{id}/deliveries/{deliveryId} (with raw_body)
- POST /api/autopilots/{id}/deliveries/{deliveryId}/replay -> creates
a new delivery row (replayed_from_delivery_id set), dispatches a
new run, never collapses onto the original via dedupe.
- PUT /api/autopilots/{id}/triggers/{triggerId}/signing-secret
Write-only; trigger response surfaces has_signing_secret +
signing_secret_hint (last 4 chars), never the secret itself.
Signature verification reuses the GitHub-compatible
X-Hub-Signature-256: sha256=<hex(hmac(body, secret))> scheme; the
HMAC helper is constant-time. Invalid/missing signatures still count
against per-IP and per-token rate limits.
autopilot_run.trigger_payload is intentionally preserved — delivery
records the HTTP receipt; run records the normalized envelope handed
to the agent. They are two different views.
Tests (Postgres-backed):
- delivery persistence on accept
- dedupe via Idempotency-Key and X-GitHub-Delivery; run_only retry
storm pin (3 retries -> 1 run)
- invalid signature: 401 + rejected row + no run linkage
- missing signature when secret configured: 401 + 'missing' state
- valid signature dispatches
- signing secret never echoed in trigger responses; hint shows last 4
- min-length and clear-by-empty for signing secret PUT
- replay creates a NEW delivery + new run; rejected deliveries cannot
be replayed
- list omits raw_body; detail includes it; cross-autopilot ID returns
404 (workspace isolation defense in depth)
- provider validation: unknown -> 400, github -> 201 round-trips
- bad-signature stream still counts against per-token rate limit
Co-authored-by: multica-agent <github@multica.ai>
* fix(autopilots): address PR review on webhook delivery layer (MUL-2334)
- Exclude `failed` from the (trigger_id, dedupe_key) partial unique index
alongside `rejected`, so a transient ingress failure does not strand the
provider's stable X-GitHub-Delivery / Idempotency-Key retry. Update the
dedupe lookup to prefer non-terminal rows under the same predicate.
- Tighten delivery status enum: drop `skipped` from the CHECK constraint
and from the handler. A run that was admission-skipped (e.g. runtime
offline) is now recorded as delivery=`dispatched` linked to the
skipped run, with the response payload carrying status=`skipped`.
Source of truth for skipped-ness is autopilot_run.status, not the
delivery row — keeps the Deliveries UI enum unambiguous.
- On dispatch error, link the (possibly non-nil) autopilot_run returned
by DispatchAutopilot to the failed delivery so Deliveries UI can
navigate to the run row for debugging.
- Slim list projection: ListWebhookDeliveriesByAutopilot no longer pulls
raw_body / selected_headers / response_body — a 100-row page × 256 KiB
would otherwise round-trip ~25 MiB from Postgres per Deliveries reload.
Detail endpoint continues to return the full row.
- Fix backend CI: TestGetDelivery_ReturnsFullPayload now decodes the
response and asserts on the parsed raw_body instead of substring-
matching against an escaped JSON string; raise the test-suite default
webhook rate limits in TestMain so the shared 192.0.2.1 IP bucket
doesn't fill across the suite and leak 429s into unrelated tests.
- Add regression coverage for the dedupe-after-failure path.
cd server && go test ./... is green locally.
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
|
||
|
Kerim Incedayi
|
9418d2a2c1 |
feat(autopilots): webhook triggers (server + CLI + UI + docs) MUL-2049 (#2348)
* feat(server): add webhook trigger DB migration + sqlc queries
Lays the foundation for webhook autopilot triggers:
- partial unique index on autopilot_trigger.webhook_token (kind=webhook only)
so the public ingress route can resolve a trigger in O(1)
- GetWebhookTriggerByToken / TouchAutopilotTriggerFiredAt /
RotateAutopilotTriggerWebhookToken / SetAutopilotTriggerWebhookToken
queries, regenerated with sqlc
* feat(server): webhook token generator + payload normalizer
Two pure helpers for the webhook autopilot work:
- generateWebhookToken: 32 random bytes -> base64-url, "awt_" prefix.
256 bits of entropy keeps brute-force off the table; the prefix makes
leaked tokens recognisable in logs.
- normalizeWebhookPayload: turns arbitrary JSON into the WebhookEnvelope
shape (event/eventPayload/request) used by trigger_payload. Header- and
body-based event inference covers GitHub, GitLab, X-Event-Type, and
caller-provided envelopes; scalar/empty/invalid bodies are rejected so
the handler can answer 400.
* feat(server): generate webhook tokens and expose rotate endpoint
- New handler.Config.PublicURL fed by MULTICA_PUBLIC_URL env so
/api/autopilots/.../triggers responses can include an absolute
webhook_url alongside the always-present webhook_path.
- CreateAutopilotTrigger now mints a webhook_token via crypto/rand
for kind=webhook and ignores cron/timezone for non-schedule kinds.
api triggers stay accepted-but-inert per PLAN.md.
- New POST /api/autopilots/{id}/triggers/{triggerId}/rotate-webhook-token
protected by the existing workspace auth group; old tokens stop
working immediately because the unique-index lookup keys on the
current row value.
* feat(server): public webhook ingress route + per-token rate limiter
- New POST /api/webhooks/autopilots/{token} route, mounted outside the
authenticated group: the path token is the credential. Workspace
context is derived from the joined autopilot row, never headers.
- Body capped at 256 KiB via http.MaxBytesReader; oversized payloads
return 413 mid-read instead of being fully buffered.
- Disabled triggers / paused / archived autopilots return
200 {"status":"ignored"} so providers stop retrying.
- Skipped-runtime dispatches surface 200 {"status":"skipped"} with the
reason from the autopilot service's pre-flight admission check.
- WebhookRateLimiter interface with sliding-window in-memory + Redis
Lua-script implementations. Default 60 req/min per token. Test
coverage on the in-memory path; Redis variant fails open on cache
errors so a Redis hiccup never blocks ingress.
- Integration tests exercise token generation, dispatch, payload
envelope persistence, GitHub-header inference, paused/disabled
short-circuits, oversized rejection, and rotate-then-old-token-404.
* feat(server): include webhook payload in create_issue description
When an autopilot run is triggered by a webhook and execution_mode is
create_issue, the agent only sees the issue body — never the run's
trigger_payload. Append a 'Webhook event:' line and a fenced JSON block
with the normalized eventPayload so the agent has the inbound context
inline. Schedule / manual runs are unchanged.
Tests cover:
- schedule path keeps existing italic note, no webhook block
- webhook path emits event line + payload block, italic before block
- non-envelope JSON falls back to raw body (defensive)
- non-webhook source with payload still gets no webhook block
* feat(core): types, API client and mutations for webhook triggers
- AutopilotRunStatus gains 'skipped' so the run-list UI handles the
admission-skipped state explicitly instead of falling through to a
generic case (the backend already emits it via MUL-1899).
- AutopilotTrigger picks up optional webhook_path / webhook_url. Both
are optional so older self-hosted servers that pre-date this change
still parse cleanly.
- buildAutopilotWebhookUrl helper composes a usable absolute URL with
the priority webhook_url > apiBaseUrl + path > origin + path > path.
Tested with seven cases covering each branch.
- ApiClient.rotateAutopilotTriggerWebhookToken posts to
/api/autopilots/{id}/triggers/{triggerId}/rotate-webhook-token; the
HTTP-contract test pins URL + method.
- useRotateAutopilotTriggerWebhookToken mutation invalidates
autopilotKeys.detail on settle, mirroring the existing trigger-mutation
pattern.
* feat(views): webhook trigger UI in Add Trigger dialog and trigger row
Add Trigger dialog gains a Schedule/Webhook segmented toggle:
- Schedule reuses TriggerConfigSection unchanged.
- Webhook hides the cron config and shows a help line; the trigger is
created with kind=webhook and the URL is generated server-side.
- Toast text differentiates schedule vs webhook on success.
TriggerRow grows a webhook branch:
- Webhook icon, kind translated via trigger_kind.
- URL shown in a truncating monospace pill, with copy + rotate
buttons. Copy uses navigator.clipboard with toast feedback; rotate
uses an AlertDialog confirm because the old URL stops working
immediately.
- api triggers render a Deprecated badge and skip URL/copy/rotate
affordances.
RunRow gains a 'skipped' RUN_VISUAL entry (muted dash) so admission-
skipped runs don't fall through to a generic case. Source label uses the
new run_source i18n key instead of capitalize.
Locales: en + zh-Hans gain run_status.skipped, run_source.*,
trigger_kind.*, trigger_row.{copy_url,rotate_url,*_confirm_*,toast_*},
add_trigger_dialog.{type_*,webhook_help,toast_added_{schedule,webhook}}.
* feat(cli): support webhook trigger creation and URL rotation
- multica autopilot trigger-add now takes --kind schedule|webhook
(default schedule for backward compatibility). For webhook it skips
--cron / --timezone validation and prints the resulting webhook URL,
preferring the server-provided webhook_url and falling back to
client.BaseURL + webhook_path.
- New multica autopilot trigger-rotate-url <autopilot-id> <trigger-id>
command for rotating the bearer URL of a webhook trigger.
* docs(autopilots): add webhook trigger guide (en + zh)
Replaces the 'Webhook and API triggers are not available yet' section
with end-to-end webhook documentation: how the URL is generated, what
payload shapes are accepted, the inferred-event rules, the bearer-secret
warning + rotate flow, status-code semantics for accepted/skipped/
ignored/4xx/5xx outcomes, and the MULTICA_PUBLIC_URL self-host
configuration.
Run history list now mentions skipped status. The 'unavailable
features' section narrows to api-kind triggers, HMAC signing, IP
allowlists, and provider presets.
* feat(views): add Schedule/Webhook toggle to the create autopilot dialog
Closes the gap where a brand-new autopilot could only be created with a
schedule trigger. The right-column config now has a Trigger section
with a segmented Schedule/Webhook control:
- Schedule keeps the existing cron/timezone UI.
- Webhook hides the cron UI and shows a help line; on submit, a
kind=webhook trigger is created right after the autopilot.
In edit mode the toggle is intentionally hidden (PLAN.md treats trigger-
type changes as delete-old + create-new, not in-place updates), but the
panel still picks the right kind based on props.triggers[0].kind so a
webhook autopilot doesn't render an irrelevant cron form.
Locales: section_trigger_kind, trigger_kind_{schedule,webhook},
section_webhook, webhook_help_{create,edit} added in en + zh-Hans.
* feat(views): show webhook URL inline after creating a webhook autopilot
After a successful create with kind=webhook, the dialog stays open and
swaps to a confirmation panel showing the freshly minted URL with a
copy button + 'Treat this URL like a password' warning + Done button.
Avoids the friction of "create the autopilot, then go find it in the
list, click in, scroll to triggers, copy URL."
Locales: dialog.webhook_created_{title,description,warning,done} added
in en + zh-Hans.
Schedule create flow is unchanged (toast + close). The success panel is
gated on the trigger returned from the create mutation, so a partial
failure (autopilot created, trigger creation errored) still falls
through to the toast_create_partial path.
* feat(views): show webhook payload in run detail dialog
The agent transcript dialog now accepts an optional headerSlot that
sits above the event list. The autopilot RunRow drops a
WebhookPayloadPreview into that slot when the run came from a webhook
and trigger_payload is non-empty.
The preview is collapsed by default (the transcript itself is the main
event), shows the inferred event name + receivedAt in the header, and
reveals the eventPayload as pretty-printed JSON with a copy button on
expand. Falls back gracefully if the row's trigger_payload doesn't
match the WebhookEnvelope shape — the whole value is shown instead so
nothing is hidden.
Closes the "agent didn't echo the payload, now I can't see what
triggered the run" gap. PLAN.md tracked this as
"Payload preview in run history" under follow-ups.
Locales: webhook_payload.{label, unknown_event, payload, content_type,
copy, copied, copied_short, copy_failed} added in en + zh-Hans.
* chore(server): wire MULTICA_PUBLIC_URL through self-host compose
Two small follow-ups split out of the webhook trigger PR:
- docker-compose.selfhost.yml passes MULTICA_PUBLIC_URL into the
backend container so a self-hosted deployment behind a real domain
gets absolute webhook URLs in the trigger response. Documented in
.env.example with the rationale for not deriving the public host
from request headers.
- Drop a duplicated 'invalid json:' prefix in the webhook ingress
400 error path. normalizeWebhookPayload already prefixes its
errors, so the handler doesn't need to re-prefix.
* fix(migrations): renumber webhook trigger migration 081 → 089 to avoid collision
The branch's 081_autopilot_webhook_triggers.{up,down}.sql collided
numerically with 081_runtime_timezone.{up,down}.sql that landed on
main, making migration apply order undefined. Renumber to 089 so the
file slots after the latest main migration (088_squad_instructions).
The SQL itself doesn't conflict — it only creates a partial unique
index on autopilot_trigger.webhook_token — but the duplicate prefix
is what the migration runner sees, so the filename must move.
* fix(autopilot-webhook): address PR review blocking issues
- Redact bearer tokens from request logs: paths matching
/api/webhooks/autopilots/<token> now log "[redacted]" instead of the
token. The resolved trigger ID is plumbed via context so audit lines
stay useful for debugging. (Review item Blocking #1.)
- Distinguish pgx.ErrNoRows from transient DB errors in token lookup:
no-row stays 404 (so providers don't retry on a deleted webhook),
other errors return 500 (which providers DO retry, avoiding silent
drops on DB blips). (Review item Blocking #2.)
- Add per-IP sliding-window rate limiter that runs BEFORE the token
lookup, so spraying random tokens can no longer probe the
autopilot_trigger index unboundedly. Reuses the existing Lua script
with a separate Redis key namespace; falls open on Redis errors.
Default budget 30 req/min/IP. (Review item Blocking #3.)
The webhook handler now applies the gates in the order: per-IP rate
limit → token lookup → per-token rate limit → handler logic.
* fix(autopilot): atomic webhook trigger creation + strict kind/timezone validation
- Mint the webhook bearer token BEFORE the INSERT and pass it via
CreateAutopilotTriggerParams so the row never exists in a half-written
kind=webhook + webhook_token=NULL state. On the (vanishingly rare)
unique-index collision the whole INSERT is retried with a fresh token
— no UPDATE second step. Removes the now-dead attachFreshWebhookToken
helper. (Review item Recommended #4.)
- Add new GET /api/autopilots/{id}/runs/{runId} endpoint that returns a
single run including the full trigger_payload. The list response is
now slim (omits trigger_payload) so worst-case payload size drops
from ~5 MB to ~5 KB. (Review item Recommended #5, server side.)
- Reject kind=api with 400 ("kind=api is deprecated; use schedule or
webhook") and reject kind=webhook with --timezone with 400 — both
surfaces stragglers loudly instead of silently dropping fields.
CLI mirrors the check so --timezone with --kind webhook errors
client-side. (Review nits.)
- Add --yes (-y) flag and an interactive y/N confirmation prompt to
`multica autopilot trigger-rotate-url` so the destructive rotate
matches the UI's AlertDialog safety. (Review item Recommended #6.)
* fix(views): fetch webhook payload on-demand and truncate at 4 KiB
- Add useAutopilotRun query hook + getAutopilotRun API client method
paired with the new server endpoint. The run-detail dialog now mounts
a WebhookPayloadSlot that fetches the full run (incl. trigger_payload)
lazily — list responses no longer carry up to 256 KiB × N runs of
envelope data.
- WebhookPayloadPreview truncates its in-DOM <pre> at 4 KiB with a
localized marker so jank-y machines aren't asked to render a 256 KiB
JSON blob. The Copy button still yields the full string.
- Adds the truncated_marker i18n string to en + zh-Hans.
Review items Recommended #5 (frontend) and a nit on the preview's
unbounded <pre>.
* test(autopilot-webhook): close coverage gaps flagged in PR review
- request_logger: redactWebhookPath unit tests + integration test
proving the bearer token never lands in slog output, plus the
webhook_trigger_id context plumbing.
- autopilot_webhook_handler: empty body → 400, archived autopilot →
200 ignored, per-IP rate limiter trips before DB lookup, kind=api
and webhook+timezone are rejected at 400, slim list + full detail
endpoint round-trip.
- webhook_rate_limiter: Lua script structure guard (catches reordering
even without a live Redis), plus live-Redis tests for both per-token
and per-IP limiters (REDIS_TEST_URL gated, matching the existing
Redis test pattern in the package).
- WebhookPayloadPreview: envelope rendering, fallback shape, and the
>4 KiB truncation path with full-payload-on-Copy guarantee.
Two branches are documented as code-review-protected rather than
covered by tests: the 500-on-DB-error path requires injecting a stub
Queries (no interface here), and the cross-workspace defense-in-depth
check is unreachable from valid SQL state.
* fix(middleware): SetWebhookTriggerID must mutate request in place
The round-1 helper returned a fresh *http.Request from WithContext, and
the webhook handler did `r = SetWebhookTriggerID(r, ...)`. That swaps
the handler's local pointer but doesn't propagate the new context back
to RequestLogger, which is still holding the original *http.Request —
so the audit line never actually included webhook_trigger_id in
production. The round-1 test happened to pass because it pre-stashed
the value on the request before calling ServeHTTP, bypassing the bug
it was meant to verify.
Switch to in-place mutation via `*r = *r.WithContext(...)` so the
wrapping middleware sees the new context after next.ServeHTTP returns,
and update the test to exercise the real call pattern (set the context
from inside the handler, assert the surrounding logger reads it).
Verified live: an accepted webhook now logs
path=/api/webhooks/autopilots/[redacted] webhook_trigger_id=<uuid>
* fix(autopilot-webhook): symmetric ErrNoRows split + trusted-proxy gate
Round-2 review (Bohan-J, PR #2348 follow-up):
- Must-fix #1: the second lookup at autopilot_webhook.go:258
(GetAutopilot after the token resolves) was folding every error into
404. A transient DB blip would tell a webhook sender "not found" and
it would never retry. Apply the same errors.Is(err, pgx.ErrNoRows)
→ 404 / else → 500 split as the first lookup got in round 1.
- Must-fix #2: clientIPForRateLimit was honoring X-Forwarded-For /
X-Real-IP from any caller. An attacker spraying random tokens could
just rotate the XFF header and the per-IP bucket became per-request,
so the limiter that's specifically supposed to gate spraying before
it hits the DB unique index was bypassed.
New shape — matches Bohan's suggestion exactly:
* Default: r.RemoteAddr only, headers ignored.
* Operator opt-in via MULTICA_TRUSTED_PROXIES (comma-separated
CIDRs). XFF/X-Real-IP are honored only when r.RemoteAddr is
inside one of the listed prefixes; otherwise they're dropped.
Wired through .env.example and docker-compose.selfhost.yml so
self-host operators can configure their reverse-proxy's CIDR.
Invalid CIDRs in the env var are dropped with a single slog.Warn at
startup rather than crashing the server. Uses net/netip (stdlib,
value-typed) for parsing and containment checks.
Verified live on the rebuilt self-host backend: a 35-request spray
from one source with rotating XFF gets the expected 30× 404 + 5× 429,
proving the per-IP bucket is keyed on the real connection IP.
* fix(autopilot): reject cron/timezone PATCH on non-schedule triggers
Round-2 review should-fix. CreateAutopilotTrigger already 400s on
kind=webhook + timezone/cron_expression, but UpdateAutopilotTrigger
silently wrote those fields regardless of prev.Kind. The values then
sat in the DB visible to nobody and read by nothing — a back door that
left the API contract fuzzy across create vs update.
Mirror the create-path discipline: after loading prev, if prev.Kind
!= "schedule" and the PATCH body sets cron_expression or timezone,
return 400 with a clear message. enabled and label remain accepted on
every kind.
The existing prev.Kind == "schedule" guard on next_run_at recompute
stays as belt-and-braces, but with this gate in place the recompute
branch is now reachable only for the kind it was meant for.
* test(autopilot-webhook): close round-2 coverage gaps
- IPRateLimitNotBypassedByXFFSpoof: drives the must-fix #2 invariant
by rotating XFF across three calls from the same RemoteAddr and
asserting the third gets 429. Pre-round-2 this test would have
passed for the wrong reason (limiter trusted XFF, so per-bucket
collision was incidental); now it pins the bypass-closed property.
- IPRateLimitReturns429BeforeDBLookup: updated to set RemoteAddr
explicitly and drop the XFF header it was leaning on. With
TrustedProxies empty (test default) the limiter keys on the real
connection IP, which is what the test wants to assert anyway.
- UpdateAutopilotTrigger_RejectsCronExpressionOnWebhookKind +
UpdateAutopilotTrigger_RejectsTimezoneOnWebhookKind: drive the
round-2 should-fix from the handler boundary.
- UpdateAutopilotTrigger_AcceptsEnabledAndLabelOnWebhookKind: counter
test so a regression to a blanket reject is caught.
* fix(migrations): bump webhook trigger migration 089 → 091
origin/main added 089_squad_no_action_activity_index (and 090_task_is_leader)
since our last rebase, re-colliding with our 089_autopilot_webhook_triggers.
Bump to 091 so the filename ordering is unambiguous again. The SQL is
unchanged — same partial unique index on autopilot_trigger.webhook_token —
only the filename moves.
* fix(views): dedupe skipped icon in autopilot RUN_VISUAL after rebase
The rebase against origin/main merged main's add of `Ban` for the
skipped status next to our round-1 `MinusCircle` entry, leaving the
RUN_VISUAL map with two `skipped` keys (only the last would have been
read at runtime, and MinusCircle had been dropped from the imports
during conflict resolution — so the file would not compile).
Keep main's `Ban` icon (latest design) and a single `skipped` entry.
Carry over the round-1 comment about why the muted styling matters
for failure-ratio readability.
---------
Co-authored-by: Kerim Incedayi <kerim.incedayi@digitalchargingsolutions.com>
|
||
|
iYuan
|
d8635ad580 |
fix(issues): prevent duplicate active issue creation (MUL-2225) (#2602)
* fix: prevent duplicate active issue creation * fix(issues): address duplicate guard review * fix(autopilot): skip duplicate issue admissions * fix(issueguard): tighten duplicate lookup edge cases * test(issues): cover duplicate guard autopilot skips * feat(autopilots): group skipped runs in history |
||
|
LinYushen
|
319b23eb39 |
Revert "feat(task): add claim lease mechanism (Phase 2, MUL-2246) (#2660)" (#2674)
This reverts commit
|
||
|
LinYushen
|
3137feecdf |
feat(task): add claim lease mechanism (Phase 2, MUL-2246) (#2660)
Add claim_token + claim_expires_at columns to agent_task_queue and three new SQL queries for the claim lease protocol: - ClaimAgentTaskWithLease: generates a UUID token and sets a lease expiry when claiming a task, so the daemon must prove it received the response - StartAgentTaskWithClaimToken: validates the token on StartTask, preventing stale daemons from starting requeued tasks - RequeueExpiredClaimLeases: moves dispatched tasks with expired leases back to queued for re-claim This closes the reliability gap where a claim response lost in transit leaves a task stuck in dispatched until the 60s dispatch timeout fires. Co-authored-by: multica-agent <github@multica.ai> |
||
|
Jiayuan Zhang
|
4d6b5ad06f |
fix(squad): wake leader when dual-role agent posts as worker (MUL-2218) (#2626)
* fix(squad): wake leader when dual-role agent posts as worker (MUL-2218) The squad-leader self-trigger guard skipped a comment whenever the author equalled the squad's leader id, regardless of the role the agent was acting in. For an agent that holds both leader and worker roles in the same squad, this meant the leader role never reacted to its own worker output and the issue stalled. Tag each enqueued task with is_leader_task and consult the agent's most recent task on the issue from both self-trigger guards (comment path + @squad mention path) — skip only when that task was itself a leader task. Co-authored-by: multica-agent <github@multica.ai> * fix(squad): inherit is_leader_task on retry task clone (MUL-2218) CreateRetryTask cloned a parent task into a fresh queued attempt but omitted is_leader_task from the column list, so the child silently fell back to the column default (false). For a leader task that hit auto-retry through MaybeRetryFailedTask, the retried task posed as a worker task — the self-trigger guard then no longer recognised the leader's own comments, re-opening the very loop MUL-2218 closes. Inherit p.is_leader_task in the clone and add a query-level test that covers both leader and worker retries. Co-authored-by: multica-agent <github@multica.ai> --------- Co-authored-by: multica-agent <github@multica.ai> |
||
|
Multica Eve
|
a2dd80d4f6 |
feat(autopilot): skip dispatch when assignee runtime is offline (MUL-1899) (#2311)
* feat(autopilot): skip dispatch when assignee runtime is offline (MUL-1899) Prevents scheduled autopilots from accumulating doomed tasks against offline / archived / unbound agents. Before this change, a paused laptop or crashed daemon would let a 5-minute-cron autopilot pile up thousands of queued agent_task_queue rows that no runtime would ever drain — this is the dominant source of the 89k stuck-task backlog flagged in MUL-1899. DispatchAutopilot now performs a pre-flight admission check on the assignee agent's runtime status. If the runtime is not 'online' (or the agent is archived / has no runtime bound / has no assignee), the run is recorded as 'skipped' with a failure_reason and no task is enqueued. Skipped runs still emit autopilot:run.done so the UI / activity feed reflect that the trigger fired and was evaluated. Skipped runs are deliberately NOT counted toward the failure-ratio auto-pause: a user who closes their laptop overnight should not have their autopilot paused. Sustained server-side failures keep their existing pause path via the failure monitor. Tests: added an integration test that creates an offline runtime and asserts DispatchAutopilot records a skipped run with no task enqueued. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Co-authored-by: multica-agent <github@multica.ai> * feat(scheduler): expire stale queued tasks via TTL sweeper (MUL-1899) Companion to the dispatch-time admission gate added in this PR. The admission gate prevents *new* tasks from being enqueued against an offline runtime, but it does not drain the historical backlog (~89k stuck queued rows observed at MUL-1899 baseline) and does not help when a runtime goes offline *after* a task has already been queued. This adds a passive TTL sweeper: - New SQL query `ExpireStaleQueuedTasks` transitions queued tasks older than the TTL to status='failed' with failure_reason='queued_expired' and a clear error message. - Sweep is capped per tick (`queuedExpireBatchSize`, default 500) via a CTE+LIMIT so that draining a large backlog cannot monopolise the DB on a single tick. At 30s ticks the worst case is 60k rows/hour. - Wired into the existing 30s `runRuntimeSweeper` loop alongside `sweepStaleTasks` and reuses `taskSvc.HandleFailedTasks` so the expired tasks broadcast `task:failed` events, reconcile agent status, and roll back any in-progress issues — same lifecycle as any other failed task. - Default TTL = 2h. Conservatively above any reasonable "queued behind a long-running task" window (default agent timeout is 2h, sweeper runs every 30s) so legitimate work isn't expired. - Integration tests cover the happy path (stale → expired, fresh → left alone, correct status/reason/error) and the per-tick batch cap. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Co-authored-by: multica-agent <github@multica.ai> * fix(autopilot): address review blockers from PR #2311 (MUL-1899) GPT-Boy review of the offline-runtime + queued-TTL PR flagged four blockers; this commit addresses them all. 1. Restore the 'skipped' autopilot_run status in the DB constraint. Migration 043 had removed 'skipped' along with the now-defunct concurrency_policy feature, so the new admission gate's INSERT of status='skipped' violated `autopilot_run_status_check` and broke `TestAutopilotDispatchSkipsWhenRuntimeOffline` in CI. New migration 079 re-adds 'skipped' to the CHECK list. The down migration migrates skipped → failed before re-tightening, mirror- ing what 043 did for the original removal. 2. Make `ExpireStaleQueuedTasks` race-safe. The CTE-then-UPDATE pattern could clobber a task that the daemon claimed between victim selection and the outer update. Two guards added: - `FOR UPDATE SKIP LOCKED` in the CTE so we never wait on a row that's currently being claimed (and never block the claim path either). - The outer UPDATE now re-checks `t.status = 'queued'` AND the TTL predicate so even if a row's lock is released after a successful claim, we cannot transition a now-dispatched/ running task to 'failed'. 3. Add a partial index for the queued-TTL sweeper. `idx_agent_task_queue_queued_created_at` on `created_at WHERE status = 'queued'` — keeps the 30s sweep query (status=queued AND created_at < ... ORDER BY created_at LIMIT 500) cheap even when historical terminal rows accumulate (~89k+ at MUL-1899 baseline). The partial predicate keeps the index tiny because only in-flight rows live in 'queued'. 4. Fix the failure-monitor denominator. `SelectAutopilotsExceedingFailureThreshold` had been counting 'skipped' toward total runs, which would have diluted the failure ratio: a 100%-failing autopilot could mask itself behind a wall of admission skips. With 'skipped' restored as a real status, the auto-pause monitor must explicitly exclude it from BOTH numerator and denominator — admission skips are neither a success nor a failure. Verified: `go test ./cmd/server/... ./internal/service/...` passes (including TestAutopilotDispatchSkipsWhenRuntimeOffline, TestExpireStaleQueuedTasks, TestExpireStaleQueuedTasksRespectsBatch Limit). `go build ./... && go vet ./...` clean. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Co-authored-by: multica-agent <github@multica.ai> * fix(migrations): split queued-task TTL index into concurrent migration Per PR #2311 review: agent_task_queue is a hot table, so building the new partial index with plain CREATE INDEX inside migration 079 would hold ACCESS EXCLUSIVE on the queue and block dispatch during deploy. The migration runner does not allow CONCURRENTLY to share a file with other statements (documented in 068), so split the index into its own single-statement file 080 — matching the existing pattern in 035 / 067 / 074 / 075 / 078. Migration 079 keeps the autopilot_run constraint change. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Co-authored-by: multica-agent <github@multica.ai> --------- Co-authored-by: Eve <eve@multica-ai.local> Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Co-authored-by: multica-agent <github@multica.ai> |
||
|
LinYushen
|
250ada1fb3 |
chore(db): drop unused agent_task_queue.last_heartbeat_at (#2212)
Drops the unused agent_task_queue.last_heartbeat_at column and removes the hot-path task heartbeat write. |
||
|
Bohan Jiang
|
38f777d0ba |
feat(autopilot): auto-pause autopilots with sustained high failure rate (#2136)
* feat(autopilot): auto-pause autopilots with sustained high failure rate Adds a background monitor that pauses any active autopilot whose recent runs are dominated by failures (defaults: ≥100 terminal runs in 7d, ≥90% failed). The monitor leaves a severity=attention inbox notification for the autopilot's creator (or the agent's owner if the autopilot was agent-created) so a human learns about the auto-pause and can fix the root cause before re-enabling. Motivated by MUL-1336 §6 #2: a single broken cron autopilot (`Registro de ls cada 5 min`, 1,475/1,476 failed in 7d) was burning ~1.5k tasks/tokens per week with no human in the loop. Tunable via AUTOPILOT_FAIL_MONITOR_{INTERVAL,LOOKBACK,MIN_RUNS,FAIL_RATIO,STARTUP_DELAY}; INTERVAL=0 disables the monitor entirely. Co-authored-by: multica-agent <github@multica.ai> * chore(autopilot): relax failure monitor defaults to daily / 50 runs Per review feedback in MUL-1339: 30-min scan was overkill — the 50-run threshold already provides multi-hour lag, and operational simplicity matters. Lowering MinRuns from 100 → 50 keeps low-frequency autopilots in scope (~7 runs/day reaches threshold within 7d window). Co-authored-by: multica-agent <github@multica.ai> --------- Co-authored-by: multica-agent <github@multica.ai> |
||
|
Bohan Jiang
|
b1345685a3 |
fix(task): rerun starts a fresh session, skip poisoned resume (#1928)
* fix(task): rerun starts a fresh session, skip poisoned resume
When a task ended in a known agent fallback ("I reached the iteration
limit and couldn't generate a summary.", "Put your final update inside
the content string. Keep it concise.") the (agent_id, issue_id) resume
lookup would still pick that session, so a manual rerun inherited the
poisoned state and reproduced the same bad output.
Two complementary guards:
1. Daemon classifies poisoned terminal output and routes it through the
blocked path with failure_reason set ('iteration_limit' /
'agent_fallback_message'). GetLastTaskSession excludes failed tasks
with those reasons, so even comment-triggered tasks no longer resume
them. Tasks that failed mid-flight (timeout, runtime_recovery, etc.)
are still resumable, preserving MUL-1128's auto-retry contract.
2. Manual rerun marks the new task force_fresh_session=true. The daemon
claim handler skips the resume lookup entirely when the flag is set,
capturing the user-intent signal that "the prior output was bad" even
when poisoned classification misses a future fallback wording.
Auto-retry of orphaned mid-flight failures (MaybeRetryFailedTask →
CreateRetryTask) does not take this path, so it keeps resuming.
Tests: classifyPoisonedOutput unit test; integration tests assert the
SQL filter excludes poisoned classifiers, RerunIssue flips the flag,
and the normal enqueue path leaves it false.
Co-authored-by: multica-agent <github@multica.ai>
* fix(daemon): cap poisoned-output matcher to short trimmed text
GPT-Boy review on MUL-1630: the previous strings.Contains match would
classify any output that quoted the marker substring — including a
review/analysis that simply discussed the marker itself. Real fallback
messages are short single-sentence affairs, so cap the candidate at
~one paragraph and trim whitespace before matching. Adds regression
tests covering a long quoting review and a marker buried in a long
real conclusion; both must stay classified as completed.
Co-authored-by: multica-agent <github@multica.ai>
* fix(migrations): rename 065 force_fresh_session → 066 to clear collision
main introduced 065_project_resources after this branch was cut, so
both files shared the 065_ prefix. The readiness check
(server/cmd/server/health.go → migrations.LatestVersion) takes the
last entry by lexical order, which is 065_project_resources, leaving
this branch's 065_force_fresh_session unguarded — a deploy that
applied project_resources but not force_fresh_session would still
report ready, and the next enqueue / rerun / claim would crash on
"column force_fresh_session does not exist".
Renaming to 066_force_fresh_session puts it strictly after
project_resources so readiness blocks until it's applied.
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
|
||
|
Naiyuan Qing
|
f745a3bbbe |
feat(agent): presence v3 + execution log + trigger summary (#1823)
* refactor(views): migrate agent/runtime/skill lists to TanStack DataTable
Replace the per-page CSS Grid + minmax(min, fr) + sticky-first-col + truncate
implementation with a TanStack Table backend rendered through a Dice UI-style
DataTable shell. Column widths are now px-based via column.size, so cells
no longer shrink or auto-truncate as the viewport narrows; when the sum of
columns exceeds the viewport, the container scrolls horizontally instead.
- Add @tanstack/react-table to the catalog (8.21.3) and wire it into
packages/ui (dep) and packages/views (peerDep).
- packages/ui: new DataTable + DataTableColumnHeader + lib/data-table.ts
(getColumnPinningStyle), adapted from Dice UI's registry. The shell
renders <table> directly (skipping shadcn's <Table> wrapper) so its own
outer overflow controls both axes — no nested overflow conflicts.
- packages/views: each list now declares ColumnDef[] with explicit
cell renderers. Row click navigates to detail via onRowClick (instead of
wrapping <tr> in <a>, which is invalid HTML); kebab dropdowns
stopPropagation so they don't trigger the row navigation.
- Drop the previous AGENT_LIST_GRID / GRID_WITH_OWNER / ROW_GRID
templates and the sticky-first-col / subgrid mechanics that came with
them. agent-list-item.tsx is removed; runtime-list.tsx and
skills-page.tsx are trimmed to thin wrappers.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* feat(agent): cap description at 255 chars (db + api + ui)
Symmetric enforcement across DB, server, and UI:
- Migration 060: pre-flight truncate of any oversize rows, then ADD
CONSTRAINT NOT VALID + VALIDATE CONSTRAINT so the new check doesn't
block writes during validation.
- Server handler validates utf8.RuneCountInString on Create/Update and
rejects over-limit input with 400.
- Front-end gets AGENT_DESCRIPTION_MAX_LENGTH in core/agents/constants
(single source of truth shared by the create dialog + edit modal +
test suite) and a CharCounter component that warns at 90% and errors
past the cap.
- Description editor moves from a 288px popover to a roomy modal.
Editor body is mounted only while the dialog is open, so the local
draft state is locked in at mount time and never reset by an external
WS update — the React-recommended replacement for the
useEffect(reset, [value]) anti-pattern.
Counted in code points everywhere (rune count / spread length /
char_length) so multibyte input agrees across all three layers.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* refactor(views): data-table polish across runtime + skill lists
Builds on the DataTable migration in
|
||
|
Naiyuan Qing
|
40cea8454d |
feat(autopilot): redesign modal — simpler schema, consistent schedule UI (#1595)
Drop priority and project_id from autopilot. project_id was never exposed in the UI and priority duplicated the agent's own task queue priority. Redesign the create/edit modal as a Runbook (left) + Configuration (right) layout. Rework the Schedule section around a single visual shell so every picker aligns pixel-for-pixel on the same row: - TimeInput (new): segmented HH:MM control adapted from openstatusHQ/time-picker, driven by keyboard (ArrowUp/Down to step, ArrowLeft/Right to jump segment, digit typing with a 2s two-digit window). Replaces <input type="time">, whose native UI broke the design system. Supports a minuteOnly variant for hourly schedules. - TimezonePicker (new): searchable Popover with a fixed-width left check slot so rows stay aligned and GMT offsets never collide with the selected indicator. - Runbook editor now lives in a bordered card, giving the placeholder an input surface instead of bare document flow. Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com> |
||
|
LinYushen
|
0b1333fb00 |
feat(server): orphan-task recovery + auto-retry + manual rerun (MUL-1128) (#1476)
* feat(server): orphan-task recovery + auto-retry + manual rerun (MUL-1128)
When the daemon process crashed mid-task the issue was stuck at
in_progress for up to 2.5h: the in-flight task timeout was the only
mechanism that ever moved the row, and the runtime heartbeat sweeper
only fires after the runtime stays offline for 45s — a quick restart
beats both windows.
This change implements the A+B plan from the issue thread:
A. lifecycle hygiene
- migration 055 adds attempt / max_attempts / parent_task_id /
failure_reason / last_heartbeat_at to agent_task_queue
- new daemon-auth endpoint POST /runtimes/{id}/recover-orphans:
daemon calls it on every register so the server fails any
dispatched/running tasks the previous process left behind
- new daemon-auth endpoint POST /tasks/{id}/session: persists the
agent's session_id + work_dir mid-flight so a crash doesn't
lose the resume pointer (claude+codex emit MessageStatus with
SessionID; daemon forwards on the first one it sees)
- FailAgentTask / FailStaleTasks / FailTasksForOfflineRuntimes
now set failure_reason ('agent_error' / 'timeout' /
'runtime_offline')
B. auto-retry with resume context
- TaskService.MaybeRetryFailedTask spawns a fresh queued attempt
carrying parent's session_id/work_dir when the failure reason
is infrastructure-shaped (timeout, runtime_offline,
runtime_recovery) and attempt < max_attempts; skips autopilot
- wired into the runtime sweeper paths and TaskService.FailTask
so the user transparently sees a new in_progress run instead of
a stuck row
- new user-auth POST /api/issues/{id}/rerun + multica issue rerun
CLI for the manual escape hatch
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* fix(server): address PR review for orphan-task recovery (MUL-1128)
Three review-must-fix items on top of the A+B implementation:
1. recover-orphans now funnels through TaskService.HandleFailedTasks,
the same shared post-failure pipeline used by the runtime sweeper.
This guarantees task:failed events are emitted, agent status is
reconciled, and issues stuck in_progress with no remaining active
task are reset to todo even when no auto-retry is created
(max_attempts exhausted, autopilot, non-retryable reason).
2. RerunIssue now uses CancelAgentTasksByIssueAndAgent, scoped to the
issue's current assignee. The previous implementation called
CancelAgentTasksByIssue, which would collateral-cancel parallel
@-mention agents on the same issue.
3. GetLastTaskSession now considers both completed and failed tasks
(mirroring GetLastChatTaskSession), ordering by the most recent
timestamp. With UpdateAgentTaskSession pinning session_id/work_dir
mid-flight, an auto-retry or manual rerun of a daemon-crash failure
now actually resumes the prior conversation context instead of
starting fresh — matching the stated B-branch behaviour.
go build / go vet pass; the existing service and agent test suites pass.
runtime_sweeper / handler integration tests require a local DB with the
055 migration (and the pre-existing 050 first_executed_at column).
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
|
||
|
Jiayuan Zhang
|
f94b0100cd |
refactor(autopilot): remove broken concurrency policies and fix multiple bugs (#1048)
Remove the concurrency_policy system (skip/queue/replace) — skip had an orphan bug that permanently blocked triggers, queue didn't actually queue, and replace didn't cancel running tasks. Every trigger now simply executes. Bug fixes: - Listener now handles in_review status (was silently ignored) - Issue deletion fails linked autopilot runs before DELETE (prevents orphans) - ComputeNextRun rejects invalid timezones instead of silent UTC fallback - dispatchCreateIssue post-commit failures now properly fail the run Reliability: - Scheduler recovers lost triggers on startup (crash recovery) - New index on autopilot_run(issue_id) for deletion lookups - Migration 043 cleans up historical orphaned/skipped/pending runs |
||
|
Jiayuan Zhang
|
d88fe2608e |
feat(autopilot): scheduled/triggered automations for AI agents (#1028)
* feat(autopilot): add scheduled/triggered automation for AI agents Introduce the Autopilot feature — recurring automations that assign work to AI agents on a schedule or manual trigger. Supports two execution modes: create_issue (creates an issue for the agent to work on) and run_only (directly enqueues an agent task without issue pollution). Backend: migration (3 tables + 2 columns), sqlc queries, AutopilotService with concurrency policies (skip/queue/replace), HTTP CRUD + trigger endpoints, background cron scheduler (30s tick), event listeners for issue→run and task→run status sync. Frontend: types, API client methods, TanStack Query hooks with optimistic mutations, realtime cache invalidation, list page with create dialog, detail page with trigger management and run history, sidebar nav + routes for both web and desktop apps. * feat(autopilot): improve UX — trigger config, edit dialog, template gallery - Replace raw cron input with friendly frequency tabs (Hourly/Daily/Weekdays/Weekly/Custom), time picker, and timezone dropdown defaulting to user's local timezone - Fix Select components showing UUIDs instead of names (Base UI render function pattern) - Add Edit button on detail page opening a unified edit dialog - Remove project/concurrency/issue-title-template from create/edit (simplify for users) - Add trigger configuration inline during autopilot creation - Add template gallery on empty state (6 step-by-step workflow templates) - Rename "Description" to "Prompt" throughout UI - Inject autopilot run timestamp into issue description for agent date awareness - Treat issue status "in_review" as run completion (fixes skip on next trigger) - Make migration idempotent with IF NOT EXISTS clauses |