Two issues from #2790's HTML inline preview work:
1. HTML source view rendered as default-colored text. lowlight emits
`.hljs-tag` / `.hljs-name` for `<...>` brackets and element names, but
content-editor.css only styled the keyword / string / attr / etc.
classes — so toggling an inline ```html``` block to "source" showed
attributes colored and everything else plain. Adds the two missing
classes in light + dark.
2. HtmlAttachmentPreview carried a "Copy code" button. An HTML attachment
is a file (view + download), not an inline source snippet. The inline
```html``` fenced block (HtmlBlockPreview) is where reading / copying
source belongs. Drops the button, its state, and the useAttachmentHtmlText
`canCopy` branch — the hook is still needed for the iframe srcDoc.
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
ActorAvatar applies bg-muted on its container regardless of whether
an image is loaded, so transparent regions of PNG/SVG avatars reveal
the grey placeholder. agent-detail-inspector also wraps ActorAvatar
in an outer bg-muted div, layering a second grey square.
Make bg-muted conditional on the fallback state in ActorAvatar, and
drop the redundant bg-muted from avatar-picker's image-loaded branch
and the two inspector wrappers. Empty-state placeholders unchanged.
A self-host operator running a fork of Multica with their own patches would
have their daemon silently upgraded to the upstream GitHub release, clobbering
the fork. Self-host setups also routinely pin to an older server, so a fresh
CLI may no longer talk to it.
Flip the default: auto-update remains opt-in on api.multica.ai and defaults to
off on any other server URL. Either side can override via
MULTICA_DAEMON_AUTO_UPDATE.
Co-authored-by: multica-agent <github@multica.ai>
The TriggerRow's outer flex uses `items-start`, which made sense back
when every trigger only had one row of content (label + maybe a cron
expression). Once #2774 added the URL action row to webhook triggers
(Copy + Rotate buttons sitting on a second line inside the inner column),
the trash button stayed pinned to the top-right of the outer flex — it
visibly floats above the URL action buttons instead of lining up with
them, which reads as a layout glitch.
Move the trash button into the URL action row for webhook triggers so
all three action buttons (Copy, Rotate, Delete) share one flex container
and align by construction. Schedule and API triggers — which have no
URL row — keep the trash button pinned top-right (their bodies are
short enough that the top corner reads as "the row's right end").
Extract a `deleteButton` const so the JSX isn't duplicated, and add the
existing `delete_dialog.confirm` i18n string as the title attribute for
consistency with the other action buttons (Copy / Rotate already have
hover titles).
No behavioural change — same click handler, same confirm dialog.
* fix(autopilots): reject unknown {{...}} tokens in issue title template (MUL-2370)
`--issue-title-template` (and the matching `issue_title_template` API
field) silently kept any placeholder other than `{{date}}` as a literal
string in the rendered issue title — `{{.TriggeredAt}}`, `{{trigger_id}}`,
`${date}`, etc. would all slip through `strings.ReplaceAll` unchanged
because the renderer only knew one token. The flag name and help text
("Template for issue titles (create_issue mode)") and the docs phrasing
("the title supports interpolation like `{{date}}`") both implied a
richer placeholder set existed.
Tightens the contract on three fronts:
- Reject any `{{...}}` token other than `{{date}}` at create/update time
with `unknown template variable %q; supported: {{date}}` — turns the
silent-on-trigger surprise into an explicit 400 the moment the user
sets the template.
- Update CLI flag help on `autopilot create --issue-title-template` and
`autopilot update --issue-title-template` to spell out that only
`{{date}}` (UTC, YYYY-MM-DD) is interpolated.
- Update `apps/docs/content/docs/autopilots{,.zh}.mdx` to drop the
"like `{{date}}`" phrasing for the single supported placeholder.
Adds service-layer tests covering `interpolateTemplate` (substitution,
empty-template fallback, no-placeholder verbatim) and
`ValidateIssueTitleTemplate` (accepts empty / plain / `{{date}}` /
`{{ date }}`; rejects Go-template, Mustache-style, future placeholders
like `{{datetime}}`, and templates that mix one valid and one invalid
token).
Expanding the placeholder set (`{{datetime}}`, `{{trigger_id}}`,
`{{trigger_source}}`) is tracked as a separate enhancement — those
need run/trigger context plumbed into the renderer, which is out of
scope for this bug fix.
Closes#2732
Co-authored-by: multica-agent <github@multica.ai>
* fix(autopilots): render {{ date }} whitespace form too (MUL-2370)
Validator permitted {{ date }} but interpolateTemplate only matched the
exact string {{date}}, so a template that passed create/update could
still emit a literal {{ date }} at trigger time — re-introducing the
silent-literal behaviour the validator was meant to remove.
Route rendering through the same regex as validation so every accepted
form is also a substituted form. Cover {{ date }} substitution in
TestInterpolateTemplate.
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
* feat(editor): HTML attachments render like images (MUL-2345 v4)
HTML attachments no longer wear the file-card chrome (icon + filename
row). They now render as a sandboxed iframe with a hover-revealed
right-top toolbar (Open / Download / Copy code), mirroring the image
attachment visual model.
- New HtmlAttachmentPreview owns the iframe + hover toolbar plus three
states (loading / success / error). Failure mode keeps the toolbar
pinned open and Open/Download enabled so the user is never stranded
without an escape hatch — Copy code disables when the text body is
unavailable.
- New AttachmentBlock thin dispatcher picks the renderer per kind:
html + attachmentId + !uploading -> HtmlAttachmentPreview, else
AttachmentCard. All three entry points (file-card NodeView, readonly
file-card, standalone AttachmentList) call AttachmentBlock, so feature
work on a new kind only touches one place.
- AttachmentCard collapses back to a pure file-card row UI: the inline
HTML iframe branch (InlineHtmlIframe + inlineHtmlEnabled +
showInlineHtml) is removed.
- AttachmentBlock added to the editor barrel export.
Sandbox/server-side defenses unchanged: sandbox="allow-scripts" (no
allow-same-origin), srcDoc, server still returns text/plain + nosniff
on the /content proxy.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Co-authored-by: multica-agent <github@multica.ai>
* test(editor): pin three entry points to AttachmentBlock HTML route (MUL-2345)
Reviewer flagged that the v4 dispatcher refactor only had tests on the
shared AttachmentBlock + HtmlAttachmentPreview; the three real call
sites at file-card.tsx:59, readonly-content.tsx:279, and
comment-card.tsx:152 had no regression coverage. Reverting any one
would silently lose the inline HTML iframe path — the exact MUL-2330
regression we're meant to be locking down.
Each new test renders the real entry point with an HTML+attachmentId
fixture and asserts the dispatched iframe (sandbox=allow-scripts,
srcdoc) shows up while the AttachmentCard chrome (filename row) does
not. FileCardView and AttachmentList are exported from their files for
direct rendering, mirroring the existing CodeBlockView test pattern.
Mutation-tested locally: temporarily flipping each site back to
<AttachmentCard> turns its corresponding test red.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>
Co-authored-by: multica-agent <github@multica.ai>
Two related overflow bugs in the Delivery detail dialog (the popover you
open from a webhook deliveries row, shipped in #2784) became obvious as
soon as a real webhook payload was exercised:
1. **Horizontal overflow: minified JSON pushed dialog off-screen.**
`CodeBlock`'s `<pre>` uses `white-space: pre` (default for the tag),
which means a single-line minified JSON body had intrinsic
min-content equal to the whole line's width. The parent grid cell
inherits the default `min-width: auto` (= min-content), so a long
body propagated all the way up and blew DialogContent past its
`max-w-2xl` cap. Headers rendered fine because they're
pretty-printed JSON with real newlines.
Fix: `min-w-0` on the CodeBlock wrapper so it can shrink below
min-content, plus `whitespace-pre-wrap break-all` on the `<pre>` so
long lines wrap (`break-all` is the only modifier that breaks
mid-token, which a minified JSON body needs because it has no
whitespace to break at).
2. **Vertical overflow: dialog grew past viewport.**
`DialogContent` had no height cap. With Raw body + Headers +
Response body + Replay button stacked vertically, anything beyond
the screen edge (notably the Replay button) became unreachable.
Fix: `max-h-[85vh] overflow-y-auto` on `DialogContent`.
Both fixes are CSS-only in one file; HMR verified.
* docs(self-host): explain loopback-only bindings + reverse proxy guidance (MUL-2360)
Follow-up to #2759, which bound all docker-compose published ports to
127.0.0.1. The self-host quickstart still told cross-machine users to
point their CLI at `http://<server-ip>:8080`, which no longer works
(and shouldn't — the default JWT_SECRET/Postgres creds must not be
reachable from the open internet).
- Add a Callout to step 1 explaining the loopback-only bindings and
linking to the new reverse-proxy step.
- Split step 5 into 5a (same machine, defaults) and 5b (cross-machine),
with a minimal Caddyfile that fronts both frontend and backend on a
single hostname (including the `/ws` route with `flush_interval -1`).
Switch the cross-machine `--server-url` example to `https://<domain>`.
- Mirror the changes in the Chinese quickstart.
- Add a header comment block to docker-compose.selfhost.yml so anyone
reading the file directly understands why services don't show up on
`0.0.0.0` and what to do about it.
Co-authored-by: multica-agent <github@multica.ai>
* docs(self-host): use nginx highlighter for Caddyfile snippet
Shiki's default bundle does not include `caddy` / `caddyfile`, so
Vercel's `pnpm build` failed with:
ShikiError: Language `caddy` is not included in this bundle.
Switch the code fence to `nginx`, which is in the default bundle and
gives near-identical visual highlighting for this snippet. No content
changes — the Caddyfile inside the block is untouched.
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
- Multi-select UI for batch importing skills from a local runtime
- Server batch-dispatches up to 10 import requests per heartbeat cycle
- WS heartbeat now reads supports_batch_import from daemon payload
instead of hardcoding true, so old daemons correctly fall back to
one-at-a-time dispatch
- Raised server pending timeout to 3min and client poll timeout to 4min
to accommodate daemons that pop only one import per 15s heartbeat
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* feat(squads): show member working status on squad detail page
Add a new GET /api/squads/{id}/members/status endpoint that returns each
member's derived working/idle/offline/unstable status, the issues each
agent is currently running, and the last observed activity timestamp.
The Squad detail page's Members tab consumes this snapshot to render a
status pill and an active-issue link next to each agent, with live
refresh wired through the existing task/agent/daemon WS events.
Human members are returned with status=null so the UI can keep them in
the same list without implying a presence signal. Archived agents stay
in the response and surface as offline rather than being filtered out.
Co-authored-by: multica-agent <github@multica.ai>
* fix(squads): address review feedback on member status endpoint
- i18n the "blocked" issue-status pill in squad members tab (was a
bare literal that failed `i18next/no-literal-string` lint).
- Treat any dispatched/running task as working, even when its
`agent_task_queue.issue_id` is NULL (chat / quick-create tasks).
The agent slot is occupied regardless of whether we can render an
issue link.
- Force `offline` for archived agents so they appear in the list
but never look like they're still on duty, matching the RFC
decision in MUL-2319.
- Include `workspaceKeys.squads` in the post-reconnect /
workspace-switch bulk invalidation so members-status recovers
after a disconnect during which task/runtime events were missed.
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
The skill name Input on the detail editor uses `bg-transparent px-0`
to render as flush, chrome-less text. The base Input component also
applies `dark:bg-input/30`, which Tailwind keeps because it lives in
the `dark:` variant. In dark mode this exposes a 30% white fill that
appears flush against the text — looking like missing left padding.
Add `dark:bg-transparent` to the className so the override wins in
both color modes.
On desktop, localDaemonId is fetched async, so on first paint the only
machines available are remotes — the existing auto-select picks the
first remote, then sticks because subsequent renders see selectedMachineId
still in the list. Result: the local Mac never gets the default focus
even though it sorts first.
Re-evaluate the default on every machines change, preferring the local
section. Honor a user pick once it's been made.
Co-authored-by: multica-agent <github@multica.ai>
* feat(editor): inline HTML attachment preview + ```html block render (MUL-2345)
* attachment-preview-modal: switch HTML iframe sandbox from "" to
"allow-scripts" so JS-driven chart libraries render. The opaque-origin
iframe still cannot touch cookies, localStorage, parent state, or
top-nav — only scripts run.
* New shared AttachmentCard wired into the three attachment surfaces
(file-card NodeView, ReadonlyContent file-card branch, comment-card
standalone AttachmentList). HTML attachments now render inline via a
sandboxed iframe pulled through the existing /content proxy; other
kinds keep the original chrome behavior.
* New HtmlBlockPreview for fenced ```html blocks in ReadonlyContent —
default preview iframe, source/Copy toggle. Two-layer code+pre unwrap
mirrors the Mermaid pattern; unwrap now matches on language-* class
because react-markdown invokes pre before the code renderer runs.
* CodeBlockView (Tiptap NodeView) renders an iframe preview for
language=html with a CSS-hidden toggle to the editable source — the
<NodeViewContent as="code"/> mount must remain in the tree.
* Shared use-attachment-html-text hook keeps inline and modal HTML
rendering on the same React Query cache.
* Vitest coverage: allow-scripts assertion, attachment-card kind
branches, readonly HTML iframe + Mermaid unwrap regression, NodeView
editable + preview/source toggle.
No backend changes; server-side text/plain + nosniff defense kept.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Co-authored-by: multica-agent <github@multica.ai>
* fix(editor): tighten attachment preview and pre unwrap gates (MUL-2345)
Addresses Reviewer REQUEST CHANGES on PR #2790:
1. URL-only text/html attachment cards no longer surface a dead Eye
button. `AttachmentCard` previously allowed preview when
`previewableFromUrl=true` regardless of kind, but the modal's
`tryOpen` rejects URL-only text kinds because the `/content` proxy
is ID-keyed. Drop the `previewableFromUrl` prop and gate the
no-attachmentId path strictly to URL-previewable media kinds
(pdf/video/audio).
2. Readonly `pre` unwrap now uses exact class-token matching. The
previous `className.includes("language-html")` check also fired
on `language-htmlbars`, silently stripping its `<pre>` wrapper.
Use `/(^|\s)language-(html|mermaid)(\s|$)/` so only the exact
tokens unwrap.
Regression tests:
- `report.html + no attachmentId` asserts no Preview button.
- `pdf URL-only` asserts Preview button still appears.
- `htmlbars` / `mermaidx` fences keep their `<pre><code>` wrapper.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>
Co-authored-by: multica-agent <github@multica.ai>
The base docker-compose.yml bound postgres to 0.0.0.0:5432 and
docker-compose.selfhost.yml bound postgres/backend/frontend without
a host_ip prefix — defaulting to 0.0.0.0 on all interfaces.
On any VPS with a public IP, these services were reachable from the
internet. Docker bypasses UFW iptables chains by default, so host-
level firewall rules on these ports had no effect.
Fix: prefix every port binding with 127.0.0.1 so services are only
reachable from the host itself. This matches the documented
DATABASE_URL (which uses localhost) and does not break any legitimate
local dev or self-host workflow — connections from the host shell,
migration scripts, and the backend container (via Docker internal
network) all continue to work unchanged.
The default Electron application menu's zoomIn/zoomOut roles do not fire
reliably on macOS — Cmd+= would zoom in but Cmd+- could not undo it, so
users got stuck at the zoomed-in level with no way back.
Move the shortcut into before-input-event so the same handler covers
every platform and every keyboard layout. preventDefault here blocks
both the renderer keydown and the menu accelerator, so there's no
double-zoom risk on macOS.
Co-authored-by: multica-agent <github@multica.ai>
The watchdog fires on a "no progress" window, so the default mainly
matters for commands that go fully silent (no outputDelta). Bumping
from 2m → 3m leaves more headroom for legitimately slow silent
commands before treating them as a dropped function_call_output, at
a modest cost to recovery latency.
MUL-2337
Co-authored-by: multica-agent <github@multica.ai>
* feat(codex): add per-exec_command watchdog to escape dropped function_call_output (MUL-2337)
Codex app-server can drop the second function_call_output when two
exec_command calls fan out in the same turn and both async-yield through
the yield_time_ms boundary (observed 2026-05-18, MUL-2334 — Trump Agent
wedged for 6+ min with no semantic activity events to drive any existing
timer). The model then waits forever for the missing output; only the
10-minute semantic inactivity timeout would eventually rescue the run.
Add a per-call watchdog in the codex client that tracks open
exec_command / commandExecution items by call_id and fails the turn
quickly (default 2 min, configurable via ExecOptions.ExecCommandStuckTimeout)
when one stays open without progress. outputDelta events reset the
per-call progress timestamp so long-running streaming commands aren't
flagged.
This is a daemon-side mitigation only — codex itself still has the
upstream race, but the daemon no longer burns the full inactivity budget
before the run is marked failed and a new run can recover.
Co-authored-by: multica-agent <github@multica.ai>
* feat(codex): track legacy exec_command_output_delta in watchdog (MUL-2337)
Mirrors the raw v2 item/commandExecution/outputDelta refresh on the legacy
codex/event protocol so a long-running streaming exec doesn't get falsely
flagged as stuck after begin + 2 min.
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
Wires the frontend onto the PR1 webhook delivery layer. Adds a Deliveries
section to the autopilot detail page that lists recent deliveries
(queued / dispatched / rejected / ignored / failed) with provider, event,
attempt count, and timestamp. Clicking a row opens a detail dialog with
raw body, headers subset, response body, signature status, and a Replay
button. Replay is disabled client-side for signature-invalid / rejected /
still-queued deliveries to mirror the server's 400.
Backend contract is locked behind a lenient zod schema via
parseWithFallback — unknown future status / signature_status values
degrade to a generic row instead of dropping the whole list.
Co-authored-by: multica-agent <github@multica.ai>
* 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>
* fix(views): surface backend error messages on mutation failures (MUL-2317)
Mutation toasts across the views package were swallowing the backend
`error` string and showing only a generic i18n fallback. This made it
impossible for users to see why an operation failed (most visibly:
creating an issue with a duplicate title produced a vague "Failed to
create issue" toast).
The fix has three pieces:
1. Create-issue duplicate branch (A段)
- New schema `DuplicateIssueErrorBodySchema` in core/api/schemas.ts.
- `create-issue.tsx` parses `ApiError.body` via `parseWithFallback`
and renders a dedicated amber-toned toast with a "view existing"
link when the server returns `{ code: "active_duplicate_issue",
issue: {...} }`. Schema drift downgrades to the normal error toast.
- Schema intentionally omits `issue.status` so the toast does not
depend on `StatusIcon`, which has no fallback for unknown enums.
2. User-facing mutation failure toasts (B段)
- 47 sites converted to `err instanceof Error && err.message ?
err.message : <existing fallback>` — preserves all existing
code-specific branches (slug conflict, agent_unavailable,
daemon_version_unsupported) and i18n keys.
- Covers Type 1 (onError) and Type 2 (catch block) patterns across
issues, projects, autopilots, inbox, runtimes, squads, comments,
batch actions, workspace create, and agent config tabs.
3. Autopilot partial-success (Type 3)
- New i18n keys `toast_create_partial_with_reason` /
`toast_update_partial_with_reason` (double-brace `{{reason}}`).
- `autopilot-dialog.tsx` captures `err.message` in the schedule
`catch` and routes to the `_with_reason` variant when present,
preserving the partial-success semantic (autopilot saved, schedule
failed) while exposing the actual reason.
Explicitly out of scope:
- `packages/core/` mutation hooks (no global onError, no UI dependency)
- No `toastApiError` helper (matches existing 14+ correct sites)
- Sub-issue link aggregate `Promise.allSettled` keeps count-based toast
(N independent requests cannot collapse to one err.message); only
added a dev-side `console.error` per rejection.
- Clipboard catches and `useUpdateChatSession` (not API mutation toasts)
Tests:
- `packages/core/api/schemas.test.ts` — schema contract (valid body,
forward-compat fields, rename rejection, missing issue, wrong types).
- `packages/views/modals/create-issue.test.tsx` — duplicate toast +
view link, schema-drift fallback, err.message surfacing, non-Error
fallback (4 new cases).
- `packages/views/autopilots/components/autopilot-dialog-i18n.test.ts`
— real i18next, asserts rendered text contains the reason verbatim
(guards against `{reason}` vs `{{reason}}` regression).
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Co-authored-by: multica-agent <github@multica.ai>
* fix(autopilots): unify rotate-token catch + cover dialog partial-success render
Address reviewer feedback on PR #2772:
1. webhook-token rotate (`autopilot-detail-page.tsx`) now follows the
`err.message ?? fallback` ternary used by the sibling trigger
delete/add paths, instead of swallowing the error.
2. Extract `formatSchedulePartialFailureToast` so the dialog's
partial-success branches and the i18n test exercise the same
helper. The test now drives the actual format function, so a
variable-name typo at the call site (e.g. `{ msg }` instead of
`{ reason }`) fails the substring assertion.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Co-authored-by: multica-agent <github@multica.ai>
* test(modals): drop user.type for title in success path to dodge CI 5s timeout
The success-path test typed the 42-character title via userEvent which
triggers a controlled re-render per keystroke. On the slower CI runner
the whole test crept up to ~5s and intermittently tripped the default
vitest timeout. Setting the value in one shot via fireEvent.change cuts
the cost while leaving the submit + toast interactions on userEvent.
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>
Co-authored-by: multica-agent <github@multica.ai>
* feat(auth): cache workspace membership for daemon heartbeat path
Cache workspace membership existence (not role) in Redis to eliminate a
DB round-trip on every PAT-authenticated daemon heartbeat. Follows the
existing PATCache nil-safe pattern.
Key design decisions per reviewer feedback:
- Cache existence only (sentinel "1"), not role string. Authorization
decisions that depend on role always hit the DB directly. This
eliminates the cache-aside race where a stale elevated role could
persist after a downgrade.
- Proactive invalidation on UpdateMember, DeleteMember, LeaveWorkspace,
and DeleteWorkspace (iterates members before cascade delete).
- 5 min TTL. Combined with PATCache (10 min), worst-case revocation
delay is max(10m, 5m) = 10 min — consistent with original PATCache
design decision.
Limitations:
- Non-members still hit DB on every request (negative caching not
implemented — the scenario is rare for daemon endpoints which require
valid workspace-scoped tokens).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Co-authored-by: multica-agent <github@multica.ai>
* test(auth): drive membership cache invalidation through real handlers
- TestRequireDaemonWorkspaceAccess_CacheHit now uses a ghost user with no
member row, so the only path to a granted access is the cache short-circuit.
Without priming the cache the access check must fail; with priming it must
succeed. A future change that bypasses the cache would fail the second
assertion.
- Replaces the cache-only InvalidatedOnMemberRemoval test (which only
re-exercised the auth-package primitive) with four handler-driven tests
that exercise DeleteMember, UpdateMember, LeaveWorkspace and
DeleteWorkspace via their real HTTP handlers. Each test prepares a real
member, primes the cache, calls the handler, and asserts the cache entry
is gone — so a refactor that drops one of the Invalidate(...) calls in
workspace.go will fail CI.
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Co-authored-by: multica-agent <github@multica.ai>
Co-authored-by: Jiang Bohan <bhjiang@outlook.com>
Adds REDIS_URL, RATE_LIMIT_AUTH, RATE_LIMIT_AUTH_VERIFY, and
RATE_LIMIT_TRUSTED_PROXIES to the environment-variables page (EN +
ZH) and to .env.example, with the reverse-proxy caveat that without
RATE_LIMIT_TRUSTED_PROXIES every user shares the proxy IP and the
whole deployment ends up in one bucket.
Follow-up to #2636. MUL-2251.
Co-authored-by: multica-agent <github@multica.ai>
Adds a Redis-backed fixed-window rate limiter middleware on /auth/send-code,
/auth/verify-code, and /auth/google. Prevents brute-force enumeration,
verification_code table flooding, and connection pool exhaustion from
rapid-fire unauthenticated requests.
Key design decisions per reviewer feedback:
- X-Forwarded-For trust model: XFF is NEVER trusted by default. Only
honored when RemoteAddr is from a CIDR in RATE_LIMIT_TRUSTED_PROXIES.
Uses rightmost-untrusted algorithm (walks XFF right-to-left, returns
first non-trusted IP). Matches the project's conservative model in
health_realtime.go.
- Atomic INCR+EXPIRE via Lua script: prevents a stuck key (permanent
ban) if EXPIRE fails independently. Follows existing Lua script
pattern in runtime_local_skills_redis_store.go.
- Fixed-window counter (not sliding-window): simple, adequate for auth
rate limiting where precision at window boundaries is acceptable.
- Fail-open with startup warning: nil Redis disables rate limiting
(same as PATCache), but logs a warning at startup so ops can see.
- IPv6 normalization: net.ParseIP().String() produces canonical form.
- Configurable via env vars: RATE_LIMIT_AUTH (default 5/min),
RATE_LIMIT_AUTH_VERIFY (default 20/min), RATE_LIMIT_TRUSTED_PROXIES.
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
* fix(editor): sync ContentEditor when defaultValue changes externally
Tiptap v3 `useEditor` reads `content` only at mount (ueberdosis/tiptap#5831
— by design), so when an issue description is updated remotely (WS event,
another agent, another client), the editor kept showing stale content
until the issue was closed and reopened. `key={id}` in issue-detail only
force-remounts on issue switch, not on same-issue updates.
Add a useEffect in ContentEditor that watches `defaultValue` and applies
it via `editor.commands.setContent()` with four guards:
1. Focused AND dirty — protect bytes the user is actively typing.
Focused-but-clean intentionally falls through: onBlur has no replay
path, so an unconditional `if (isFocused) return` would drop the
sync forever for users who click into the editor without typing.
2. Unfocused AND dirty — covers the blur → debounce (1500ms) window
where the editor holds unsaved content but isFocused is already
false. The pending onUpdate flush reconciles via the cache;
overwriting here would be silent data loss.
3. Normalized-equal short-circuit — avoids a no-op transaction when
the cache reflects a write this editor just emitted.
4. `emitUpdate: false` — Tiptap v3 flipped setContent's emitUpdate
default to true; without this the sync would re-trigger onUpdate
→ server save → self-write loop.
After setContent, clamp the prior selection to the new doc size so the
caret doesn't snap to position 0.
Tests cover five cases: unfocused+dirty-content (sync fires),
focused+dirty (skip), focused+clean (must sync — regression guard for
the focused-but-clean hole), unfocused+dirty (blur-before-debounce
window, skip), and normalized-equal short-circuit (skip).
Closes#2409
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Co-authored-by: multica-agent <github@multica.ai>
* test(editor): cover normalized-equal sync path with a distinct defaultValue
The previous rerender passed the same `defaultValue` string, so React's
dep-array equality short-circuited the sync effect entirely — the test
only exercised the first-mount equality check, not the actual
normalized-equal guard.
Pass a different-but-trimEnd-equivalent value so the effect re-runs and
the normalized-equal short-circuit is what keeps setContent uncalled.
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>
Co-authored-by: multica-agent <github@multica.ai>
Trim the default runtime brief Available Commands to the agreed core set, including issue create/update, while keeping non-core commands discoverable through help. CI passed for backend and frontend.
* 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>
compactDeviceInfo was flipping the parenthetical of an agent CLI version
string (e.g. "2.1.5 (Claude Code)" -> "Claude Code 2.1.5") and using that
as the per-machine subtitle. Each daemon's runtimes are sorted alphabetically
and `claude` always sorts first, so every claude-equipped machine's row
ended up showing "Claude Code …" — drowning out actual per-machine differences.
The reshape was meant for OS+arch shapes ("macOS (x86_64)" -> "x86_64 macOS"),
not version strings. Filter agent-version-like parts out before picking a
primary so the subtitle either reflects real machine info or falls back to
the daemon-id descriptor.
Co-authored-by: multica-agent <github@multica.ai>
Follow-up to #2716. Updates two stale comments that still described
openclaw's `name` and `id` as interchangeable. The actual contract:
`id` is the routing key passed to `openclaw agent --agent <id>`;
`name` is a human display label and is not safe to pass to the CLI.
No behavior change.
Co-authored-by: multica-agent <github@multica.ai>
openclawEntriesToModels() used the agent Name (which may contain
spaces, e.g. "Sub2API OPS") as Model.ID. This ID is passed to
openclaw via --agent, where normalizeAgentId mangles spaces into
hyphens ("sub2api-ops"), causing a lookup miss against the
registered id ("sub2api") and a "no parseable output" error.
Fix: prefer agent ID for Model.ID; use Name only for display Label.
When ID is empty, fall back to Name for backward compatibility.
Fixes#2714
* feat(github): mirror PR CI checks and merge conflict status (MUL-2228)
Surface "checks passed/failed" and "conflicts/no conflicts" badges under
each linked PR on the issue page so users can judge readiness without
flipping over to GitHub. CI state is fed by check_suite webhooks
(GitHub Actions + apps using the Checks API; legacy status events are
out of scope for MVP); conflicts are read from pull_request.mergeable_state.
Data model:
* github_pull_request: add head_sha + mergeable_state
* github_pull_request_check_suite: per-suite rows keyed by (pr_id, suite_id)
* Aggregation done at query time, filtering by current head_sha so
late-arriving suites for a stale head can't contaminate the new head's
pending view; per-app latest suite chosen first so a single app firing
multiple suites isn't counted N times.
Webhook hardening:
* synchronize/opened/reopened/edited(base) explicitly clear mergeable_state
* single-row ordering protection on the check_suite upsert prevents a
late-delivered older event from overwriting a newer one
* check_suite.pull_requests is iterated; unknown PRs are logged and dropped
UI:
* PR row shows Checks + Conflicts badges; opaque mergeable values
(blocked/behind/unstable/...) render as no badge, not as conflicts.
* Terminal PR states (merged/closed) suppress the status row entirely.
Tests: * Pure unit coverage for derivePRMergeableState + aggregateChecksConclusion
* Webhook integration tests: multi-app aggregation, old-head ignore,
late-older-event ignore, synchronize clears mergeable_state
* Vitest coverage for pull-request-list badge rendering across CI/conflict
combinations and the legacy (null) fallback.
Co-authored-by: multica-agent <github@multica.ai>
* fix(github): scope check_suite PR lookup; preserve mergeable on metadata
Addresses code review on PR #2632.
1. check_suite handler now resolves the PR through the workspace-scoped
GetGitHubPullRequest query instead of GetGitHubPullRequestByRepoNumber.
The (workspace_id, repo_owner, repo_name, pr_number) tuple is the real
uniqueness key, so a bare (owner, repo, number) lookup could return a
stale row from another workspace and either land the suite on the wrong
PR or skip the right one when the installation ids drifted. The old
unscoped query is removed.
2. derivePRMergeableState now returns (value, clear) and the upsert SQL
distinguishes three cases: state-changing actions clear the column to
NULL, non-empty payloads write the value, and metadata events with an
empty payload preserve the existing column. Previously every empty
payload became NULL, so a labeled/assigned event silently wiped a
known clean/dirty verdict in violation of the RFC's "metadata empty
payload preserves" rule.
3. ListPullRequestsByIssue narrows to the issue's PR ids before running
the per-app check_suite aggregation, avoiding a full-table scan over
github_pull_request_check_suite when only a handful of rows belong to
the requested issue.
New helper test covers labeled+empty preserves; new integration test
verifies a metadata event after a known mergeable_state keeps the value.
Co-authored-by: multica-agent <github@multica.ai>
* feat(github): PR card layout v3 increment — stats + segmented progress bar
Replaces the row + badge layout under "Pull requests" on the issue
detail sidebar with a card that mirrors the GitHub PR summary look:
title, author/avatar, +N −M · K files diff stats, segmented progress
bar (failed → pending → passed, failure leftmost), and a one-line
status caption following an explicit priority pass-through.
Backend
- Migration 092: github_pull_request adds additions / deletions /
changed_files (INT NOT NULL DEFAULT 0). Zero defaults are what the
new frontend treats as "legacy backend — hide the stats row" so old
PR rows that pre-date this migration don't render "+0 −0 · 0 files".
- pull_request webhook handler reads stats off the top-level payload.
- ListPullRequestsByIssue now surfaces per-suite counts
(checks_passed / failed / pending) alongside the existing aggregate
conclusion, so the segmented bar reuses the already-computed counts
with no new aggregation.
Frontend (packages)
- core/github/pull-request-status.{ts,test.ts}: pure-function module
for the status-kind priority table and the segment derivation; 15
cases covered, includes the "all-zero → hide stats" guard.
- views/issues/components/pull-request-list.tsx: PullRequestCard plus
a compact-row fallback used when count > 4 (first 3 as cards, the
remainder collapsed behind a Show more toggle).
- i18n: new `pull_request_card_*` keys in en + zh-Hans.
Tests
- 12 component tests covering each rule of the priority table, the
legacy-zero stats fallback, and the collapse threshold.
- Reuse of the v3 webhook handler tests confirmed.
Verification
- pnpm typecheck + pnpm test green (60 test files, 536 tests).
- go build ./... + go vet ./... clean.
- 6 demo issues (DEV-2..DEV-7) screenshotted via Playwright; see the
PR comments for the visual check matrix.
Co-authored-by: multica-agent <github@multica.ai>
* fix(views): collapse PR cards at N>=4, not N>4
The card-vs-collapse threshold used `>` so 4 PRs slipped past it and
all rendered as full cards, contrary to RFC v3 (N >= 4 collapses to
3 cards + compact tail). Switch to `>=` and update the threshold-
boundary test to expect "Show 1 more".
Co-authored-by: multica-agent <github@multica.ai>
* fix(views): align PR sidebar rows with existing list style
Co-authored-by: multica-agent <github@multica.ai>
* fix(views): hide terminal PR status badges
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
Local daemon previously logged mostly at Info, leaving startup/exit,
config resolution, registration, heartbeat ticks, agent invocation, and
result classification undiagnosable without code-reading. Add Debug
logs at those checkpoints so LOG_LEVEL=debug (the default) produces
enough detail to follow a run end-to-end without changing normal Info
output.
Co-authored-by: multica-agent <github@multica.ai>
* fix(daemon): bump idle watchdog default 5m → 30m (MUL-2300)
The previous 5 min default killed legitimate long assistant outputs (e.g.
RFC-length writeups) where the model streams a single message for many
minutes without any daemon-visible activity. 30 min keeps the safety net
for truly stuck runs (dockerd hang) while leaving headroom for long
writes.
runIdleWatchdog tick interval is window/2, with a 30 s floor that only
applies when interval < 30 s — at window=30 min the natural tick is 15
min, so no sync needed.
Co-authored-by: multica-agent <github@multica.ai>
* docs(daemon): drop stale 5-minute mention from idle watchdog comment
Refers to DefaultAgentIdleWatchdog so the comment stays in sync if the
default shifts again. Follow-up to Emacs review on PR #2728.
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
The issue-detail right-rail Token usage card is fed by useQuery(issueUsageOptions(id)),
but the realtime task: handler only invalidated ["issues","tasks"]. As a result the
card only refreshed on remount, so consecutive runs on the same issue left the
numbers stuck until the user navigated away and back. Mirror the existing tasks
invalidation with a prefix invalidation of ["issues","usage"] so any task
lifecycle event refreshes the aggregated usage numbers.
Co-authored-by: multica-agent <github@multica.ai>
Before: dispatchCreateIssue copied autopilot.created_by_type/id onto the
new issue's creator_type/creator_id, and the same fields were used as the
ActorType/ActorID of the issue:created event. Result: any issue spawned by
an autopilot was reported as created by the human who first configured
the autopilot, not by the agent that actually owns the work. Downstream
subscriber/activity/notification listeners inherited the same wrong actor.
After: creator and actor are both the autopilot's assignee agent
(creator_type=agent, creator_id=ap.assignee_id). The human owner is still
recoverable via origin_type=autopilot + origin_id.
Audited the other ap.created_by_* usages: analytics attribution
(autopilotActorID, task.go user-id), and the private-agent visibility
gate in shouldSkipDispatch — all correctly read the autopilot's owner,
not the executor, so they stay as-is.
Co-authored-by: multica-agent <github@multica.ai>
Extends the workspace /usage page Daily tokens chart toggle from
Tokens | Cost to Tokens | Cost | Time | Tasks, so users see daily
run-time and task-count trends alongside spend without leaving the page.
- New SQL `ListDashboardRunTimeDaily`: per-date totals from
agent_task_queue (terminal tasks only), scoped to workspace and
optionally project. Same time anchor as ListDashboardAgentRunTime
so day boundaries line up.
- New handler GET /api/dashboard/runtime/daily + TanStack Query option.
- New DailyTimeChart (single-series, smart h/m/s unit) and
DailyTasksChart (completed + failed stacked).
- Empty-state is per-metric so a workspace with tokens but no terminal
runs (or vice-versa) doesn't get a false "no data".
- i18n: en + zh-Hans daily.metric_time / metric_tasks + titles.
Co-authored-by: multica-agent <github@multica.ai>
* feat(views): show Total in daily token/cost chart tooltips (MUL-2282)
Add a Total row at the bottom of the daily-tokens-chart and daily-cost-chart
tooltips so users can see the precise stack sum on hover, in addition to the
per-stack breakdown.
Implemented by extending shared ChartTooltipContent with an optional `footer`
prop (ReactNode | (payload) => ReactNode) that renders below the items with a
top divider; backwards-compatible (no behavior change when footer is omitted).
Co-authored-by: multica-agent <github@multica.ai>
* fix(views): i18n Total label in chart tooltips (MUL-2282)
Lint rule i18next/no-literal-string flagged the hardcoded "Total" string
in daily-cost-chart and daily-tokens-chart tooltips. Move it to
runtimes.charts.tooltip_total and read via useT.
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
* feat(daemon): force-stop hung agent runs via idle watchdog (MUL-2281)
A backend whose subprocess hangs on a stuck child process (e.g. claude
blocked on `docker ps` against a frozen dockerd) keeps the daemon's run
record at status="running" until the full DefaultAgentTimeout (2 h)
expires, because cmd.Wait() never returns and Session.Result is never
written. MUL-2225 spent 17+ minutes in this state in the wild.
Add a per-task idle watchdog around executeAndDrain:
- Wrap the caller's ctx so a single cancel propagates to the agent
subprocess (via the ctx passed to backend.Execute) AND the drain loop.
- Stamp lastActivityAt every time the drain loop receives a message.
- Tick at window/2; when idle_for >= window AND session.Messages buffer
is empty, set a fired flag and call cancel.
- Tag the resulting Result.Status as "idle_watchdog" so runTask routes
it through a dedicated failure_reason instead of "agent_error".
Default window is 5 min, configurable via MULTICA_AGENT_IDLE_WATCHDOG;
set to 0 to disable. Tests cover the activity-then-silence case, the
zero-message case, the disabled case, and the happy path.
Co-authored-by: multica-agent <github@multica.ai>
* fix(daemon): skip idle watchdog while a tool call is in flight
A legitimate long-running tool call (npm install, docker build, test
suite) can sit silent between tool_use and tool_result for many minutes.
Without this gate, the watchdog would yank the agent mid-build.
Track unmatched tool_use messages in an atomic counter; only let the
watchdog fire when the counter is zero. tool_result clamps non-negative
so a stray result with no matching use can't re-arm the watchdog one
call too early.
Adds two regression tests:
- DoesNotFireDuringInFlightToolCall: tool_use -> silence past
window -> tool_result -> completed (must NOT fire)
- FiresAfterToolResultIfBackendStaysSilent: tool_use -> tool_result
-> silence past window (MUST fire — backend really is stuck)
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
* feat(daemon): auto-update CLI when idle (MUL-2100)
Add a periodic poller that checks GitHub for a newer multica release
every hour and self-updates when the daemon is idle, reusing the same
brew-or-download upgrade path the Runtimes-page "Update" button already
runs.
- Refactor handleUpdate to call a shared runUpdate(target) helper so
both server-triggered and auto-triggered upgrades go through the same
brew detection + atomic replace + restart.
- New autoUpdateLoop gates each tick on: opt-out flag, Desktop launch
source, dev-build version, an in-flight update, and active tasks. The
idle gate guarantees we never interrupt a running agent — busy ticks
silently retry at the next interval.
- Config: MULTICA_DAEMON_AUTO_UPDATE=false to disable (also via
--no-auto-update), MULTICA_DAEMON_AUTO_UPDATE_INTERVAL to retune the
poll period.
- IsNewerVersion / IsReleaseVersion helpers in the cli package, with
tests covering patch/minor/major bumps, dev-describe strings, and
malformed input.
- Daemon-side tests cover every skip path (updating, active tasks,
fetch failure, no-newer) plus the success path that fires
triggerRestart while keeping the updating flag held to the end.
Co-authored-by: multica-agent <github@multica.ai>
* fix(daemon): close idle race + verify checksum in auto-update (MUL-2100)
Two issues raised in PR #2679 review:
1. The first idle check in tryAutoUpdate only ran before the release-metadata
fetch, so a poller that won the claim race during the fetch could end up
handing handleTask a task that triggerRestart was about to cancel via root-
ctx cancellation. Add a strict claim barrier: runRuntimePoller now
tryEnterClaim()s before ClaimTask, and tryAutoUpdate flips pauseClaims
under claimMu only after observing claimsInFlight + activeTasks == 0.
Pollers that were already mid-claim hold claimsInFlight > 0, so the barrier
refuses to engage and the update defers to the next tick.
2. The direct-download path replaced the running binary with whatever bytes
GitHub returned, without checking checksums.txt. Pull the manifest first,
buffer the archive, and reject on SHA-256 mismatch before extraction. The
GoReleaser config already publishes checksums.txt; we just consume it.
Also tighten parseReleaseVersion so it stops accepting dev-describe shapes
like "v0.1.13-5-gabcdef0" through the patch trim, matching its docstring.
The auto-update loop already guards on IsReleaseVersion, but the lenient
parser was a footgun and the existing test name even said "not newer" while
asserting the opposite.
Tests:
- TestTryAutoUpdate_DefersWhenClaimInFlightAtBarrier (new race coverage)
- TestTryAutoUpdate_HoldsBarrierAcrossRestart / ReleasesBarrierOnUpgradeFailure
- TestTryEnterClaim_RespectsBarrier
- TestFindChecksumManifestAsset / TestParseChecksumManifest / TestVerifyAssetSHA256
- TestIsNewerVersion: dev-describe cases now expect false (matches docstring)
Co-authored-by: multica-agent <github@multica.ai>
* chore(daemon): default auto-update poll interval to 6h (MUL-2100)
1h was overly chatty for a release that lands at most a few times a week.
Operators who want a different cadence can still set
MULTICA_DAEMON_AUTO_UPDATE_INTERVAL or --auto-update-interval.
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
* feat(views): progressive disclosure for issue sidebar properties (MUL-2275)
Split sidebar Properties into a core group that always renders
(status / priority / assignee / labels) and an optional group
(due_date / project / parent) that only appears when the issue has
the value set or the user explicitly added it via a new
"+ Add property" picker. A field cleared in-session stays visible
to avoid row flicker; navigating to a different issue reseeds
visibility from that issue's set fields. The standalone "Parent
issue" card is folded into Properties as one of those optional
rows. Adds `defaultOpen` to DueDatePicker / ProjectPicker so a
newly-added row drops the user straight into edit state.
Co-authored-by: multica-agent <github@multica.ai>
* refactor(views): swap sidebar optional set to due_date + labels
Per design feedback: status / priority / assignee / project / parent
are all required and should always render in the sidebar; only
due_date and labels are progressive-disclosure optionals. Move project
and parent rows out of the optional block (drop their +Add property
menu entries and the parent special-case in addOptionalProp). Move
labels into the optional block, gated on the issue's actual attached-
label count (queried via issueLabelsOptions), with defaultOpen wired
through LabelPicker so picking "Labels" from +Add property drops the
user straight into the picker. Tests updated for the new split.
Co-authored-by: multica-agent <github@multica.ai>
* refactor(views): restore standalone parent card, move priority to optional
Parent goes back to its own collapsible section, rendered only when the
issue actually has a parent — matching the pre-MUL-2275 behavior. It is
no longer interleaved with Properties rows.
Priority joins the progressive-disclosure set (priority / due_date /
labels). New issues default to priority "none", so the row is hidden
until set or added via "+ Add property", and PriorityPicker gains
defaultOpen so the field drops straight into edit state when chosen
from the add-property menu.
Co-authored-by: multica-agent <github@multica.ai>
* refactor(issue-detail): tighten Add-property popover visual rhythm
Picked up a small visual inconsistency while reviewing the PR's UI:
the "Add property" dropdown floated above the inspector at a noticeably
larger type scale than the property rows, and each item was bare text
while the rows it sat above all rendered with an icon + value pair.
Tweaks:
- Items: `text-sm py-1.5` → `text-xs py-1`, matching the inspector
row typography and trimming row-to-row gap from 12px to 8px.
- Each option leads with the icon the resulting picker uses
(`PriorityIcon` bars / `CalendarDays` / `Tag`) so the dropdown reads
as a preview of what will appear in the new PropRow.
- Focus indicator: replace the default thick focus ring with
`focus-visible:bg-accent + outline-none`, matching the hover state
language — keyboard focus and mouse hover now look the same.
- Popover width: `w-48` → `w-44` since the labels are short and the
visual is now denser; still leaves room for translated strings.
* fix(issue-detail): dismiss Add-property popover when an option is picked
Base UI's `Popover` doesn't auto-dismiss when a child is clicked (it's
not a Menu primitive), so picking an option left the "+ Add property"
popover sitting behind the picker that auto-opens for the newly added
row — two popovers visibly stacked.
Make the Popover controlled with a local `addPropPopoverOpen` state and
close it inside `addOptionalProp` right after enqueuing the row's
auto-open. The picker still pops on mount via `defaultOpen={autoOpenProp
=== key}`, so the user flow is unchanged from their perspective:
Click "+ Add property" → menu opens
Click an option → menu closes AND target picker opens
(Was the same flow on paper before; just had the orphan popover behind
the picker.)
---------
Co-authored-by: multica-agent <github@multica.ai>
Both create dialogs were too wide at 5xl (1024px). Align with the
codebase convention for full create dialogs (create-project,
create-issue expanded) which use max-w-4xl (896px). Keeps both
modals consistent.
Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>
Co-authored-by: multica-agent <github@multica.ai>
* feat(views): refine navigation progress bar with brand color and glow (MUL-2269)
The previous 1px bg-primary bar read as near-black on light theme and
snapped on/off in a single frame, which felt abrupt despite being a small
visual element. Switch to a 2px brand-colored sweep with right-edge glow,
slower 1.4s cubic-bezier easing, and a 200ms fade-out so completion
doesn't pop.
- Container: h-px → h-0.5 (2px); always mounted with opacity-driven fade
- Bar: bg-primary → bg-brand + two-layer box-shadow glow via color-mix
- Keyframe: 1.1s ease-in-out → 1.4s cubic-bezier(0.4, 0, 0.2, 1)
Zero new design tokens (reuses existing --brand) and zero tailwind config
changes. Desktop unaffected — same component, same prefetch=no-op path.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Co-authored-by: multica-agent <github@multica.ai>
* fix(views): unmount nav progress sweep when hidden (MUL-2269)
Hiding the bar with opacity-0 left the inner element's `infinite` keyframe
animation running on every dashboard page, defeating the perceived-perf goal.
Mount the sweep only while navigating, plus the 200ms fade tail (unmount on
opacity transitionend), so nothing animates while hidden.
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>
Co-authored-by: multica-agent <github@multica.ai>
* docs(squad): address plan-review feedback for archive + role plan
Resolve the 4 items the reviewer raised on MUL-2265:
1. TS schema: declare `active_issue_count` as optional (`number | null | undefined`)
so list/create/update Squad responses don't lie about their shape; only
`getSquad` parses through SquadSchema.
2. Archive semantics: restrict TransferSquadAssignees to active issues
(status NOT IN done, cancelled) so dialog count and SQL operate on one set
and terminal-state issues keep their historical assignee.
3. Index assumption: corrected — `idx_issue_assignee (assignee_type,
assignee_id)` exists and is sufficient at realistic squad cardinality;
no new index needed.
4. Fixed `*int64` test comparison and added `.loose()` to SquadSchema per
the local schemas.ts convention.
Co-authored-by: multica-agent <github@multica.ai>
* docs(squad): plan v3 — revert to count-all/transfer-all on archive
Reviewer round 2 surfaced two structural problems with plan v2's
active-only carve-out:
1. useActorName resolves squad names via ListSquads, which filters
archived_at IS NULL. A closed issue with an archived-squad assignee
would render as "Unknown Squad".
2. The status-only update path in UpdateIssue skips validateAssigneePair,
so a done/cancelled issue with an archived-squad assignee could be
reopened to in_progress, violating the "no active issue on an archived
squad" invariant enforced elsewhere.
Both problems disappear by reverting to count-all + transfer-all: after
ArchiveSquad runs, no issue points at the archived squad, so neither
case can occur. The product trade-off is that closed historical issues
now show the leader agent instead of the archived squad in their
"Assigned to" badge — consistent with existing agent-level reassignment
behavior elsewhere in the product.
Field rename: active_issue_count -> issue_count.
TransferSquadAssignees SQL is unchanged (already transfers all).
Co-authored-by: multica-agent <github@multica.ai>
* docs(squad): add Task 2b — wrap DeleteSquad transfer + archive in one tx
Reviewer round-3 flagged that the v3 invariant ("after archive no
issue points to the squad") was asserted on the happy path only.
DeleteSquad's current best-effort impl breaks it two ways:
- transfer failure → slog.Warn but archive proceeds (Unknown Squad,
reopen-into-archived-squad bugs reappear)
- archive failure after a committed transfer → 500 with squad still
active but emptied
Task 2b rewrites DeleteSquad to run TransferSquadAssignees +
ArchiveSquad inside one pgx tx, mirroring the project.go:266-314
pattern. Publish moves below Commit. Adds two regression tests that
lock both partial-write failure modes.
Co-authored-by: multica-agent <github@multica.ai>
* feat(squad): replace native confirm() with AlertDialog and rewrite role editor as combobox
Backend:
- Add CountIssuesForSquad sqlc query (counts every issue assigned to a squad,
no status filter — matches the existing transfer-all archive semantics).
- Extend SquadResponse with optional `issue_count` (`*int64` + omitempty,
populated only by GetSquad to avoid an N+1 in the list endpoint).
- Wrap DeleteSquad's transfer + archive in a single pgx transaction so the
v3 invariant ("after archive, no issue points to the squad") is durable
rather than best-effort. Promote slog.Warn to slog.Error and check the
parseUUIDOrBadRequest ok flag (silent zero-UUID was a #1661-class latent
bug). Publish only after Commit so realtime never sees rolled-back state.
- Tests cover happy path (count, transfer-all including terminal statuses)
and both rollback directions (transfer fail / archive fail) via a
fault-injecting tx wrapper.
Frontend:
- Extend Squad TS type with `issue_count?: number | null` (optional —
list/create/update legitimately omit it). Add SquadSchema with `.loose()`
and wrap getSquad with parseWithFallback so older servers and count-error
responses degrade to the dialog's "no count" copy variant.
- Replace `window.confirm()` with shadcn `ArchiveSquadConfirmDialog`
(destructive variant, leader name + count + closed-issue caveat in the
copy, Loader2 while pending). i18n keys added under squads.archive_dialog.
- Rewrite RoleEditor as a Popover + Command combobox: Pencil affordance is
always visible, suggestions aggregate other members' roles, commit only
on Enter or selecting a suggestion (blur discards), per-member savingId
drives Loader2 so the spinner only renders on the row being saved.
Co-authored-by: multica-agent <github@multica.ai>
* fix(squad): discard RoleEditor draft on close and no-op blank Enter
Two reviewer findings on e0d754bf:
1. Closing the Popover (outside click, Esc, trigger re-click) left `query`
in state, so reopening + Enter would commit the stale draft. Clear
`query` on every non-saving close path.
2. With an existing role, opening the editor and pressing Enter on an
empty input committed "" — `commit` only no-op'd when trimmed matched
value. Treat blank Enter as a no-op; clearing a role would need an
explicit clear action that doesn't exist yet.
Add two regression tests:
- close (via outside click) → reopen surfaces a clean input; Enter does
not commit the stale draft
- blank Enter on an existing role does not call onSave
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Co-authored-by: multica-agent <github@multica.ai>
* fix(squad): add explicit Clear button to RoleEditor
Role is optional, but the previous fix turned blank Enter into a no-op
without exposing any other way to clear an existing role — that broke a
valid terminal state. Keep blank Enter as no-op; add a "Clear role"
button at the bottom of the popover that only renders when value is
non-empty and routes through onSave("").
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>
Internal navigation on web feels laggy because clicking a sidebar link blocks
0.2–0.6s with zero visual feedback — no prefetch, no Suspense fallback in the
dashboard segment, and no React transition to mark the route commit as pending.
This change adds the three pieces App Router needs to make the click→commit
window feel instant, scoped to the (dashboard) segment so auth/landing keep
their existing chrome:
- NavigationAdapter gains an optional prefetch(path). The web adapter wires
it to router.prefetch; desktop leaves it undefined (react-router has no
equivalent and doesn't need one). AppLink prefetches on hover/focus and
preserves caller-supplied onMouseEnter/onFocus/onClick.
- NavigationProvider wraps push/replace in useTransition and exposes the
pending flag via useIsNavigating(). Every useNavigation().push caller —
sidebar AppLink, command palette, post-create modal jumps — picks this up
automatically.
- New apps/web/app/[workspaceSlug]/(dashboard)/loading.tsx renders a minimal
skeleton during cold transitions inside the dashboard segment only.
- DashboardLayout renders a 1px top progress bar driven by useIsNavigating.
packages/views remains free of next/* imports; desktop is unaffected by
construction (no prefetch, transition flips quickly, no loading.tsx).
Co-authored-by: multica-agent <github@multica.ai>
* feat(task): wire claim lease queries into TaskService and sweeper (MUL-2246)
- ClaimTask now uses ClaimAgentTaskWithLease (generates claim_token + lease)
- StartTask accepts optional claim_token for token-verified start
- AgentTaskResponse includes claim_token for daemon to use
- Daemon client sends claim_token in StartTask body
- Sweeper calls RequeueExpiredClaimLeases each tick
- Legacy daemons without claim_token still work (graceful fallback)
Co-authored-by: multica-agent <github@multica.ai>
* fix(task): address PR #2662 review blockers (MUL-2246)
1. ClaimAgentTaskForRuntime: push runtime_id into atomic SQL WHERE clause
so runtime A cannot claim tasks queued for runtime B under the same agent.
2. Legacy StartAgentTask: add claim_token IS NULL guard so leased rows
cannot be started without token verification. Handler rejects malformed
tokens with 400 instead of silently degrading to legacy path.
3. StartAgentTaskWithClaimToken: validate claim_expires_at >= now(),
preserve claim_token until terminal state (only clear claim_expires_at),
use CTE + UNION ALL for idempotent retry when daemon resends after a
lost StartTask response. Return 409 Conflict on token mismatch/expiry.
Co-authored-by: multica-agent <github@multica.ai>
* fix(daemon): StartTask 409 handling, transport retry, claim_token on FailTask (MUL-2246)
- StartTask 409 (claim superseded): release slot, don't call FailTask
- StartTask transport timeout/5xx: retry once with same token, then
check task status before failing
- FailTask now sends claim_token; server-side FailAgentTask SQL adds
AND (claim_token IS NULL OR claim_token = @claim_token) guard so
stale daemons cannot fail tasks that have been re-claimed
Co-authored-by: multica-agent <github@multica.ai>
* fix(task): close FailTask token bypass and RequeueExpiredClaimLeases liveness gap (MUL-2246)
Blocker 1 - FailTask token validation:
- SQL: change (param IS NULL OR claim_token = param) to
(param IS NULL AND claim_token IS NULL) OR claim_token = param
so tokenless requests can only fail legacy (tokenless) rows.
- task.go: malformed claim_token now returns ErrInvalidClaimToken (400)
instead of being silently dropped to NULL.
- Handler: maps ErrInvalidClaimToken→400, ErrClaimTokenInvalid→409.
- Service: when UPDATE returns no rows but task is still active,
return ErrClaimTokenInvalid (token mismatch) instead of silent success.
Blocker 2 - RequeueExpiredClaimLeases runtime liveness:
- SQL: JOIN agent_runtime, only requeue tasks where runtime is 'online'.
Dead/offline runtime tasks stay dispatched for FailTasksForOfflineRuntimes.
- FOR UPDATE → FOR UPDATE OF atq (required with JOIN).
Regression tests:
- task_claim_token_test.go: malformed, tokenless-on-tokened, wrong-token
- requeue_lease_test.go: SQL must JOIN agent_runtime with online filter
Co-authored-by: multica-agent <github@multica.ai>
* fix(task): move expired lease requeue to ClaimTaskForRuntime preflight, add heartbeat freshness backstop (MUL-2246)
- Add RequeueExpiredClaimLeasesForRuntime: per-runtime preflight self-requeue
in ClaimTaskForRuntime. Runtime proves liveness by actively claiming, so no
heartbeat check needed.
- Update global RequeueExpiredClaimLeases to require ar.last_seen_at freshness
(stale_threshold_secs param). Prevents requeuing to a dead runtime in the
90s gap between lease expiry (60s) and offline detection (150s).
- Add regression tests verifying the heartbeat freshness check and that the
preflight query does not join agent_runtime.
Co-authored-by: multica-agent <github@multica.ai>
* fix(task): use LivenessStore for global requeue, move preflight before empty-cache (MUL-2246)
Blocker 1: Global RequeueExpiredClaimLeases now uses LivenessStore.IsAliveBatch
to verify runtimes are truly alive before requeuing expired leases. When
LivenessStore is unavailable (no Redis), global requeue is skipped entirely —
the preflight self-requeue in ClaimTaskForRuntime handles live runtimes. This
closes the 60-150s gap where a dead runtime still appears online in DB.
Blocker 2: Moved RequeueExpiredClaimLeasesForRuntime BEFORE EmptyClaim.IsEmpty
fast-path in ClaimTaskForRuntime. Expired leases are now requeued (which bumps
the empty cache via notifyTaskAvailable) before the empty check can
short-circuit the claim path.
Also adds ListRuntimesWithExpiredClaimLeases SQL query and LivenessChecker
interface on TaskService.
Co-authored-by: multica-agent <github@multica.ai>
* fix(task): wire EmptyClaimCache into backend taskSvc for backstop requeue (MUL-2246)
The backend taskSvc used by the sweeper only had Liveness wired but not
EmptyClaim. When global backstop requeue called notifyTaskAvailable,
s.EmptyClaim.Bump() was a nil no-op — the handler's empty-cache was never
invalidated, so the daemon's next claim hit a stale empty verdict.
Fix: wire the same Redis-backed EmptyClaimCache into the backend taskSvc
in main.go (same Redis keys as router.go:139 handler instance).
Add regression test verifying backstop requeue invalidates the handler's
empty-cache.
Co-authored-by: multica-agent <github@multica.ai>
* fix(task): global backstop must not requeue — alive runtimes use preflight, dead stay dispatched (MUL-2246)
- RequeueExpiredClaimLeases is now a no-op (returns 0 always)
- Alive runtimes self-requeue via ClaimTaskForRuntime preflight
- Dead runtimes stay dispatched for FailTasksForOfflineRuntimes
- Rewriting to queued on dead runtime creates 2h blackhole (offline
sweeper only handles dispatched/running)
- Test actually calls RequeueExpiredClaimLeases and asserts 0 in all cases
Co-authored-by: multica-agent <github@multica.ai>
* fix(daemon): remove duplicate usage reporting block after merge conflict (MUL-2246)
The merge resolution introduced a second ReportTaskUsage call after the
status check, duplicating the usage-before-early-return block that already
runs right after runner.run. Remove the duplicate and add a regression test
asserting /usage is called exactly once on the normal completion path.
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
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>
Each consecutive run of activities renders as a single "N activities"
summary by default. Clicking expands the block in place. Comments are
unaffected; the most recent activity block stays expanded so users see
"what just happened" without a click.
Refs MUL-2188
Co-authored-by: multica-agent <github@multica.ai>
Co-authored-by: J <j@multica.ai>
* docs(email): clarify 888888 is opt-in via MULTICA_DEV_VERIFICATION_CODE; document SMTP option in self-host docs
The startup log line, .env.example, and SELF_HOSTING_ADVANCED.md still
implied that the dev master code 888888 is auto-active whenever
APP_ENV != "production". That has not been true since the master code
was gated behind MULTICA_DEV_VERIFICATION_CODE — the fixed code is
disabled by default and must be opted in explicitly.
Also extend the docs site with the SMTP relay backend added in #1877:
auth-setup, environment-variables, and self-host-quickstart now cover
both Resend and SMTP options in EN and ZH.
Co-authored-by: multica-agent <github@multica.ai>
* docs(email): treat SMTP as an email backend in self-host docs and startup warning
Address review feedback on #2666:
- server: startup warning now fires only when both RESEND_API_KEY and SMTP_HOST
are empty, since either one is a valid email backend. Otherwise the log
mis-tells SMTP-only operators that verification codes go to stdout.
- self-host-quickstart (EN/ZH): tell readers to fetch the verification code
from whichever backend they configured (Resend or SMTP); fall back to
stdout only when neither is configured.
- auth-setup (EN/ZH): \"without Resend\" → \"without any email backend
configured\" so the wording stays correct now that SMTP is a first-class
option.
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
* fix(realtime): include actor_type in WebSocket broadcast messages
The WS broadcast message format was {type, payload, actor_id} but missing
actor_type. This meant the web UI could not distinguish agent from human
operations in real-time events at the top level.
While payload data for comments (author_type) and activities (entry.actor_type)
already included the type, the top-level message did not — causing the web UI
to display agent CLI operations as human operations when relying on the
broadcast actor identity.
Changes:
- server/cmd/server/listeners.go: add actor_type to all broadcast messages
- packages/core/types/events.ts: add actor_type to WSMessage interface
- packages/core/api/ws-client.ts: pass actor_type to event handlers
- packages/core/realtime/hooks.ts: update EventHandler type signature
- packages/core/realtime/provider.tsx: update EventHandler type signature
Fixes MUL-2260
Co-authored-by: multica-agent <github@multica.ai>
* test: add frame-shape unit test asserting actor_type in WS frames
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
* fix(deps): refresh pnpm-lock.yaml after #2665 added test deps to core
#2665 (MUL-2256, fix(realtime)) added `@testing-library/react` and
`react-dom` to `packages/core/package.json` devDependencies, plus moved
`react` from dependencies → devDependencies, but didn't commit the
regenerated lockfile. CI runs `pnpm install` with --frozen-lockfile
(implicit in CI envs), which bails immediately:
ERR_PNPM_OUTDATED_LOCKFILE: pnpm-lock.yaml is not up to date with
packages/core/package.json
* 2 dependencies were added: @testing-library/react@catalog:,
react-dom@catalog:
Frontend CI has been red on main since 7c8cf929. Backend is fine
because Go doesn't share the lockfile.
Lockfile delta is small (+9 / -3): the only changes are the three
specifier blocks for the deps already declared in package.json. No
version upgrades, no transitive churn — `pnpm install` produced an
identical resolved tree minus the missing entries.
* fix(core): name the test wrapper component to satisfy react/display-name
Same source of CI red as the lockfile bump in this PR — #2665 also
introduced packages/core/realtime/use-realtime-sync-ws-instance.test.tsx
where `createWrapper` returned an anonymous arrow component. The
`react/display-name` lint rule (enforced as error in core) flagged it,
and once `pnpm install` was unblocked the next CI step fell through to
this lint failure.
Convert the inline arrow into a named `function Wrapper(...)` —
identical render output, satisfies the rule.
Verified: `pnpm --filter @multica/core lint` → 0 errors (was 1).
The 4 tests in this file still pass.
* fix(realtime): invalidate workspace queries on WSClient instance change
When switching workspaces, the old WSClient is torn down and a new one
is created. Events emitted during the transition are lost because
onReconnect only fires for reconnections within the same instance.
Add an effect that tracks the WSClient instance via useRef and, on
detecting a non-initial new instance, invalidates all workspace-scoped
queries (same set as onReconnect). The first assignment is skipped to
avoid redundant refetches on initial mount.
Closes multica-ai/multica#2562
Co-authored-by: multica-agent <github@multica.ai>
* refactor(realtime): extract shared invalidation helper + add ws instance test
- Extract invalidateWorkspaceScopedQueries() to deduplicate the
invalidation key list shared by onReconnect and ws-instance-change effects
- Add hook test covering: first ws skip, null gap no-op, new instance
invalidates exactly once, same instance no re-invalidation
Addresses review nits from PR #2665.
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
* feat(email): add SMTP relay as alternative to Resend
Self-hosted deployments often run behind a corporate firewall with an
existing SMTP relay (Exchange, Postfix, sendmail) and no access to
external SaaS APIs. Resend requires a public domain, an API key, and
outbound HTTPS to api.resend.com — all unavailable in air-gapped or
private-network setups.
This adds a second email delivery path using Go's stdlib net/smtp,
activated when SMTP_HOST is set. Priority order:
1. SMTP relay (SMTP_HOST set)
2. Resend API (RESEND_API_KEY set)
3. DEV stdout (neither set)
New env vars (all optional, no breaking change):
SMTP_HOST — SMTP server hostname
SMTP_PORT — port, default 25
SMTP_USERNAME — for authenticated SMTP; empty = unauthenticated relay
SMTP_PASSWORD — used only when SMTP_USERNAME is set
SMTP_TLS_INSECURE — set to "true" to skip TLS cert verification
(for private CA / self-signed certs)
The implementation:
- Dials TCP, creates smtp.Client manually (avoids smtp.SendMail which
does not expose TLS config)
- Tries STARTTLS if advertised; uses InsecureSkipVerify only when
SMTP_TLS_INSECURE=true (opt-in, nolint:gosec annotated)
- Applies PlainAuth only when SMTP_USERNAME is non-empty
- Wraps all errors with context for easier debugging
- Reuses existing HTML templates from buildInvitationParams for
invitation emails (no template duplication)
Also updates .env.example and docker-compose.selfhost.yml with the
new variables and inline documentation.
* fix(email): add dial timeout, session deadline, RFC headers for SMTP path
Address review blockers from multica-eve and Bohan-J (PR #1877):
- net.Dial → net.DialTimeout(10s) + conn.SetDeadline(30s) so a blackholed
SMTP relay cannot hang SendVerificationCode (called synchronously from the
auth handler) or leak goroutines in the invitation path.
- Add Date, Message-ID, and proper Content-Transfer-Encoding headers.
Date is required by RFC 5322; many strict relays reject messages without it.
Message-ID aids deliverability and threading.
- MIME-encode Subject via mime.QEncoding so non-ASCII workspace/inviter names
(CJK, emoji) survive without corruption across any RFC 2047-conformant relay.
- Probe 8BITMIME after (possible) STARTTLS: use Content-Transfer-Encoding 8bit
when the relay advertises 8BITMIME, quoted-printable otherwise — safe for
all relay configurations without forcing base64 overhead.
- Update SELF_HOSTING_ADVANCED.md to document Option B (SMTP relay) alongside
the existing Resend section, including all five env vars and a note that
port 465/SMTPS is not yet supported.
* fix(email): correct has8Bit assignment order (bool is first return of Extension)
handleTask had two early-return paths that ran before ReportTaskUsage:
the cancelledByPoll select and the post-run GetTaskStatus check. Both
silently discarded any usage accumulated by the agent — and both
claude.go and codex.go populate Result.Usage even when runCtx is
cancelled mid-run, so cancelled tasks consistently under-reported tokens.
Hoist ReportTaskUsage to run immediately after the runner returns,
before any early-return path. Add a taskRunner interface seam and a
cancelPollInterval field so tests can inject a fake runner and trigger
the poll-cancellation path on a 10ms ticker without spawning real agents.
Two regression tests cover both leak windows:
- TestHandleTask_ReportsUsageBeforeCancel: post-run /status returns
"cancelled"; usage must be reported before the status check.
- TestHandleTask_ReportsUsageWhenCancelledByPoll: poll goroutine fires
first and cancels runCtx; runner returns usage on Done; assert
poll-status precedes usage (proving the cancelledByPoll branch was
the one exercised, not the post-run path).
Sanity-checked: reverting only the ReportTaskUsage hoist fails both
tests with the original "tokens lost" message.
MUL-2258
Co-authored-by: Jiang Bohan <bhjiang@outlook.com>
Co-authored-by: multica-agent <github@multica.ai>
Without the full [@Name](mention://<type>/<UUID>) syntax, the platform
does not trigger the target agent. Add an explicit, strongly-worded
hard rule at the top of the list so the leader model never forgets.
Co-authored-by: multica-agent <github@multica.ai>
* feat(squad): accept avatar_url on CreateSquad
Threads avatar_url through the SQL query, sqlc-generated code, and the Go
handler so the create-squad flow can persist an avatar at creation time
instead of forcing a follow-up PATCH.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Co-authored-by: multica-agent <github@multica.ai>
* feat(squad): add avatar_url to CreateSquadRequest
Extends the TS contract for the new backend field so the frontend can pass
an uploaded avatar URL through api.createSquad.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Co-authored-by: multica-agent <github@multica.ai>
* feat(squads): rework Create Squad modal to match CreateAgentDialog (MUL-2233)
Replaces the cramped small-dialog flow with the same large-dialog shape used
by Create Agent: identity row (AvatarPicker + name + description with char
counter), grouped Leader picker (My Agents first, then Workspace Agents),
and a new multi-select Additional Members picker covering agents and
workspace members. The members trigger collapses to "+N" once more than
three are selected; promoting an agent to leader auto-drops it from the
additional-members list.
After createSquad, additional members are attached via Promise.allSettled
so a single failure surfaces a warning toast without blocking navigation —
the squad still exists and the user can retry from the Members tab.
Adds packages/views/modals/create-squad.test.tsx covering identity binding,
leader-group ordering, leader/member conflict sanitization, the empty- and
partial-failure success paths, and the create-failure recovery path.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Co-authored-by: multica-agent <github@multica.ai>
* fix(squads): valid trigger HTML + drop conflicted leader from members
Two issues from PR #2645 review:
1. AdditionalMembersPicker's PopoverTrigger was a <button> containing
MemberChip's remove <button>, which React/HTML flags as nested
interactive content (hydration + a11y warning). Render the trigger as
a <div role="combobox"> via Base UI's render prop so the chip's
remove button is valid.
2. sanitizedMembers only hid the leader from rendered/submitted output,
so promoting an additional member to leader then switching leader
away resurrected the hidden pick. Drop it from selectedMembers at
the moment of promotion via handleLeaderChange; sanitizedMembers is
no longer needed.
Adds a test that promotes → switches leader and asserts the member is
not resubmitted.
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>
Co-authored-by: multica-agent <github@multica.ai>
Backend now validates http/https/ssh/git scheme plus scp-like
`git@host:owner/repo.git` shorthand, but three repo URL inputs were
still `type="url"`. The browser's native URL validation rejected scp
shorthand with "Please enter a URL" before the value could reach the
backend.
- Switch the three inputs to `type="text"` so submission isn't blocked
client-side (project resources picker, workspace repositories tab,
create-project repo picker).
- Extend the en/zh placeholders to show a scp shorthand example
alongside the existing https one.
- Add a repositories-tab test that types `git@github.com:...` and
asserts the input is text-type, passes native validity, and reaches
the update mutation.
Co-authored-by: multica-agent <github@multica.ai>
* fix(projects): accept SSH repo URLs for github_repo resources (#2484)
The project resource validator rejected anything that wasn't http(s), so
workspace repos configured with an SSH remote (ssh:// or the scp-like
`git@host:owner/repo.git` shorthand) could not be attached to a project.
Both forms are valid git remotes and the daemon hands the URL straight to
`git clone`, so the API has no reason to require https specifically.
Relax the validator to accept http/https/ssh/git schemes and the scp-like
shorthand, while still rejecting pasted garbage (no scheme, missing host,
missing path, ftp://, file://, etc.).
Co-authored-by: multica-agent <github@multica.ai>
* fix(projects): reject scp-like URLs with '@' after ':' to avoid panic
isValidGitRepoURL indexed '@' and ':' independently, then sliced
s[at+1 : colon]. For inputs without '://' where '@' appears after the
first ':' (e.g. `host:org/repo@branch`), `at+1 > colon` triggered a
slice-bounds panic instead of a 400. Guard the slice and treat such
inputs as malformed.
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
* fix(daemon): disable Claude AskUserQuestion in non-interactive mode (MUL-2244)
GitHub #2588: when Claude Code calls its built-in AskUserQuestion tool
inside the daemon's stream-json runtime, the question never reaches the
user — there's no UI to render it — so the SDK returns an empty answer
and the agent silently "infers" and continues. From the issue's
perspective, execution looks stuck while the agent is actually charging
ahead on its own guess.
Two-part fix:
- `buildClaudeArgs` now passes `--disallowedTools AskUserQuestion` so
the tool is not exposed to the model at all.
- The Claude-specific runtime brief tells the agent to use a `blocked`
issue comment for genuine clarification, or to state an explicit
assumption and proceed.
Adds a regression test that pins both: AskUserQuestion is forbidden in
CLAUDE.md and is NOT mentioned in the AGENTS.md emitted for non-Claude
providers (the tool is Claude-specific).
Co-authored-by: multica-agent <github@multica.ai>
* refactor(daemon): drop CLAUDE.md AskUserQuestion guidance, rely on --disallowedTools
The --disallowedTools flag already prevents Claude from invoking
AskUserQuestion, so duplicating the rule in the runtime brief just bloats
the prompt without changing behavior. Removes the section and its
regression test; the argv-level test in pkg/agent already pins the flag.
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
Adds a regression test for `anthropic/claude-opus-4.7-20251001` that
exercises all three resolvePricing tolerances at once (provider strip,
Claude dot→dash, date trim). Each step was already covered pairwise;
this nails down their composition so a future change to candidate
ordering can't silently drop a step.
Follow-up to #2654 (MUL-2243); raised in second review.
Co-authored-by: multica-agent <github@multica.ai>
Copilot's `meta.agentMeta.model` reports Claude SKUs with dots
(`claude-opus-4.7`, `claude-sonnet-4.6`, ...), and openclaw / opencode
emit the `<provider>/<model>` form (`anthropic/claude-opus-4.7`). The
maintained MODEL_PRICING table only keys on Anthropic's canonical
dashed form (`claude-opus-4-7`), so every Copilot-routed turn was
falling through to the "Custom model pricing" dialog and silently
contributing $0 to cost totals.
Teach `resolvePricing` two new tolerances, in order before date stripping:
1. Strip a leading `<provider>/` segment — that's routing metadata,
not part of the SKU.
2. For `claude-*` IDs only, normalize dots to dashes. Scoped to
Anthropic because for OpenAI the separator is semantic (`gpt-5.4`
is a distinct SKU from a hypothetical `gpt-5-4`).
Custom pricing still wins over nothing, but the maintained catalog
still wins over a stale custom override (existing invariant preserved
by the test suite).
Co-authored-by: multica-agent <github@multica.ai>
The chat header dropdown was capped at max-w-80 while the trigger
could grow unbounded with the current chat title, so the popup
appeared narrower than the trigger and titles inside were truncated
early. Cap the trigger at max-w-96 and let the popup inherit the
trigger width via --anchor-width with the same upper bound, so the
two stay visually consistent and only truncate at extreme lengths.
Co-authored-by: multica-agent <github@multica.ai>
Sidebar "新建 issue" button, command palette "New Issue", and the `c`
shortcut all hard-coded which create modal to open, ignoring the
persisted lastMode in useCreateModeStore. Pressing `c` after switching
from agent → manual reverted to agent on the next open.
Add `openCreateIssueWithPreference(data?)` helper next to the store.
Generic entries call it; entries that pre-seed manual-only fields
(status, project_id, parent_issue_id from board / list / project /
sub-issue actions) keep opening "create-issue" directly because agent
mode does not honour those seeds.
Co-authored-by: multica-agent <github@multica.ai>
* feat(desktop): silent background auto-download for updates (MUL-2224)
Flip electron-updater to autoDownload=true so new releases are pulled in
the background without user action; the UI now only surfaces a
"ready to install" prompt once the package is fully downloaded.
- updater.ts: autoDownload=true; update-downloaded forwards version +
releaseNotes; single-flight guard around checkForUpdates() so startup,
periodic, and manual triggers don't pile up overlapping downloads.
- preload: update-downloaded payload now carries { version, releaseNotes? }.
- update-notification.tsx: drop available/downloading UI; ready state has
Later / Restart now and renders the version from the download event.
- updates-settings-tab.tsx: settings copy now describes background download
+ restart prompt instead of a download prompt.
Co-authored-by: multica-agent <github@multica.ai>
* fix(desktop): swallow unhandled downloadPromise rejection in updater (MUL-2224)
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
* feat(execenv): native OpenClaw skill discovery via per-task config
MUL-2213 stopped lying about native discovery and routed openclaw skills
to .agent_context/skills/ — a path openclaw's scanner never reads.
Multica skills attached to openclaw-backed agents were still invisible to
the runtime; the AGENTS.md fallback was only a documentation patch.
OpenClaw's skill scanner walks <workspaceDir>/skills/ (plus a few other
roots), and workspaceDir is resolved from the openclaw config file —
specifically agents.list[id].workspace → agents.defaults.workspace →
~/.openclaw/workspace. There is no CLI flag or env var override on the
agent runtime; the only knob is the config file.
This change wires a per-task synthesized config:
1. execenv.prepareOpenclawConfig deep-copies the user's existing
openclaw.json (priority: $OPENCLAW_CONFIG_PATH, else
~/.openclaw/openclaw.json), rewrites agents.defaults.workspace AND
every agents.list[].workspace to the task workdir, and writes the
result to {envRoot}/openclaw-config.json. Provider sections,
registered agents, model providers, gateway settings — everything
openclaw needs to actually start — are preserved as-is.
2. resolveSkillsDir for "openclaw" now points at {workDir}/skills/,
which is the first path openclaw scans under workspaceDir. Skills
written here are picked up natively.
3. daemon.go exports OPENCLAW_CONFIG_PATH={env.OpenclawConfigPath} on
the openclaw subprocess and adds OPENCLAW_CONFIG_PATH to the
custom_env blocklist so users cannot accidentally override it.
4. buildMetaSkillContent now lists openclaw alongside the
"discovered automatically" providers; the .agent_context/skills/
fallback line stays for gemini/hermes.
The new regression test TestPrepareOpenclawSkillWriteMatchesScanPath is
the one MUL-2219's DoD calls out: it resolves the workspaceDir the way
openclaw does (reading agents.defaults.workspace out of the synthesized
config) and proves {workspaceDir}/skills/<name>/SKILL.md is what Multica
actually wrote. The pre-MUL-2219 fix asserted "we wrote a file" without
checking the scanner would ever see it — which is how the dead drop into
.openclaw/skills/ landed in #2621's first commit.
Verified locally: minimum-viable synthesized config validates via
`openclaw config validate`, and `OPENCLAW_CONFIG_PATH=<path> openclaw
config get agents.defaults.workspace` returns the task workdir as
expected. MUL-2219
Co-authored-by: multica-agent <github@multica.ai>
* fix(execenv): delegate openclaw config parsing to CLI and fail closed
Address Elon's must-fix on PR #2628: the previous implementation parsed
~/.openclaw/openclaw.json with encoding/json, which cannot read JSON5
or follow $include — the OpenClaw spec's actual format. When parsing
failed, prepareOpenclawConfig silently emitted a minimal config, which
could boot OpenClaw without the user's registered agents, model
providers, or API keys.
Two changes:
1. Delegate active-config-path resolution and config reading to the
openclaw CLI itself. `openclaw config file` locates the active
config (covering OPENCLAW_CONFIG_PATH / OPENCLAW_STATE_DIR /
OPENCLAW_HOME / default and the legacy chain), and the wrapper we
write uses $include to point at it so OpenClaw's own loader handles
JSON5, $include nesting, env-substitution, and secret refs. We read
only agents.list via `openclaw config get --json` to rewrite each
entry's workspace — secrets, comments, and includes in the user
config are never touched.
2. Remove the silent minimal-config fallback. Any CLI failure,
malformed output, or write error now surfaces as a hard error from
Prepare / Reuse. The only "synthesize minimal" path left is a fresh
install (CLI reports a path but the file doesn't exist), where
there is no user data to lose.
The per-task override still rewrites every agents.list[].workspace,
not just agents.defaults.workspace — this is intentional task
isolation, documented in prepareOpenclawConfig and the PR body. A
host-scope per-agent workspace would otherwise silently route the
scanner back to the user's shared workspace.
Cleanups Elon flagged in the same review:
- daemon.go inline-system-prompt comment no longer claims openclaw
ignores the task workdir; it does load it now, and the inline brief
is a belt-and-suspenders carryover for older releases.
- execenv.go openclaw block no longer references "skill file paths in
the inline brief" — the brief uses "discovered automatically".
Reuse() switches to a ReuseParams struct so the openclaw binary path
threads through alongside CodexVersion without a 6th positional arg.
MUL-2219
Co-authored-by: multica-agent <github@multica.ai>
* fix(execenv): grant OpenClaw $include cross-dir confinement for per-task wrapper
The per-task wrapper at envRoot/openclaw-config.json $includes the user's
active config (typically ~/.openclaw/openclaw.json), but OpenClaw confines
$include resolution to the wrapper file's directory unless the target's
parent is granted via OPENCLAW_INCLUDE_ROOTS. Without this, OpenClaw refuses
to follow the link at runtime and the wrapper boots with no user-registered
agents.
prepareOpenclawConfig now returns dirname(activePath) as IncludeRoot, and
the daemon prepends it to whatever the user already has in
OPENCLAW_INCLUDE_ROOTS via the new composeOpenclawIncludeRoots helper
(dedupes, drops empty segments, preserves user-configured roots). Fresh
install emits no $include and leaves the env var untouched.
Adds OPENCLAW_INCLUDE_ROOTS to the custom_env blocklist so a per-agent
override cannot strip the granted root.
Regression tests:
- TestPrepareOpenclawConfigWrapperLoadableUnderIncludeConfinement asserts
every $include target's dirname is covered by the IncludeRoot we surface.
- TestPrepareEnvironmentOpenclawWiresIncludeRoot covers the non-fresh-install
Environment wiring.
- TestComposeOpenclawIncludeRoots covers the daemon-side env composition
(preserve, dedupe, drop empties).
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
The RUNTIME cell rendered base name + (hostname) with both spans using
flex: 0 1 auto, so the longer hostname dominated and squashed the name
to a single letter. Give the base name shrink priority and let the
hostname own the flex slot with basis-0, so hostname truncates first
while the name stays readable.
Co-authored-by: multica-agent <github@multica.ai>
* 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>
* MUL-2215: fix(daemon): close handleRuntimeGone success/straggler race
handleRuntimeGone coalesced concurrent recoveries with a per-workspace
`reregisterNextAttempt` slot that was deleted immediately on success. A
late-arriving goroutine whose `removeStaleRuntime` was delayed by mutex
contention could reach the coalesce gate after the winner cleared the
slot, observe no slot, re-claim, and double-register — the source of the
intermittent `register endpoint called 2 times under stampede, want 1`
failure on PR #2348.
The slot delete on success is intentional (a genuinely later distinct
deletion in the same workspace must register again, validated by
TestHandleRuntimeGone_DistinctDeletionsWithinCoalesceWindowBothRecover),
so we can't just extend the slot's lifetime.
Add a second per-workspace gate: `reregisterLastCompletedAt`. Every call
captures `entryAt` at the top of handleRuntimeGone; at the coalesce gate
a caller bails if `lastCompletedAt >= entryAt`, i.e. a peer's register
completed AFTER we entered the function. Same-wave stragglers bail
deterministically; distinct later events have `entryAt > lastCompletedAt`
and proceed.
Extracted the gate into `tryClaimRegisterSlot` / `recordRegisterCompletion`
so the race can be exercised deterministically with synthetic timestamps
instead of relying on `-count=N` to win the scheduling lottery.
- TestHandleRuntimeGone_CoalescesConcurrentCallers: -count=500 -race
clean (previously intermittent).
- New unit tests cover the straggler bail, the distinct-later-event
claim, failure backoff suppression, and peer-holds-slot coalescing.
Co-authored-by: multica-agent <github@multica.ai>
* MUL-2215: narrow completion stamp to success path
Second review caught that recordRegisterCompletion stamped
lastCompletedAt on both success and failure. A failed register has not
covered any workspace state, so a same-wave straggler whose entryAt
predates the failure must be allowed to retry once the failure backoff
expires — the previous behavior would let the failure-time stamp also
hide that straggler. workspaceSyncLoop only retries when a workspace's
runtimeIDs fully drain, so partial-deletion recovery has to come from
the straggler path.
Failure path now only updates reregisterNextAttempt; success path keeps
its existing stamp + slot clear. Add a regression test covering the
entryAt-before-failed-completion / arrival-past-backoff edge.
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
* fix(execenv): write OpenClaw skills to .openclaw/skills/ for native discovery
The OpenClaw provider was missing a case in resolveSkillsDir, so workspace
skills attached to OpenClaw-backed agents fell through to .agent_context/
skills/ — a path the openclaw CLI never inspects. The result: agents
created against the OpenClaw runtime saw zero of their loaded Skills in
chat or task runs, even though the meta AGENTS.md content advertised
them as auto-discovered.
Mirrors the same per-provider mapping already in place for OpenCode,
Copilot, Pi, Cursor, Kimi, Kiro. Also adds .openclaw to the repocache
git-exclude list so the per-task skills directory does not pollute
checked-out repos. MUL-2213
Co-authored-by: multica-agent <github@multica.ai>
* fix(execenv): drop .openclaw/skills dead-drop write; flag openclaw as non-auto-discovery
Reviewer (Elon) pointed out that {workDir}/.openclaw/skills/ is not in any
OpenClaw skill discovery path. Confirmed by reading openclaw upstream
(src/agents/skills/refresh.ts, src/agents/agent-scope-config.ts,
src/cli/program/register.agent.ts):
- OpenClaw scans <workspaceDir>/skills, <workspaceDir>/.agents/skills,
~/.openclaw/skills, ~/.agents/skills, bundled, and config
skills.load.extraDirs.
- workspaceDir is resolved from the openclaw config (per-agent
workspace -> agents.defaults.workspace -> ~/.openclaw/workspace).
It is NOT the cwd of the openclaw process.
- There is no --workspace CLI flag on 'openclaw agent', and no
OPENCLAW_WORKSPACE env var consumed at runtime. The only knob is the
config file.
So {workDir}/.openclaw/skills/ written by Multica is never seen by the
openclaw runtime, and the meta AGENTS.md was lying to the agent by
claiming auto-discovery. Reverts:
- resolveSkillsDir: drop the openclaw case; falls back to
.agent_context/skills/ (same path as hermes).
- agentGitExcludePatterns: drop .openclaw; nothing is written there now.
Also updates the openclaw branch in buildMetaSkillContent to point the
agent at .agent_context/skills/ explicitly (alongside gemini/hermes), so
loaded skills are at least referenced by path in the AGENTS.md context.
The openclaw native loader still won't see them as installed skills.
Native auto-discovery for openclaw needs per-task workspace integration
(e.g. synthesized per-task config via OPENCLAW_CONFIG_PATH that overrides
agents.defaults.workspace, or resolving the agent's actual configured
workspace at exec time) — tracked as follow-up. MUL-2213
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
* MUL-2216: feat(agents,squads): persist Mine/All tab selection per workspace
Tab selection on the Agents and Squads list pages was held in
component-local state, so navigating into a detail page and back
remounted the list and reset the tab to the default "Mine". Move
`scope` into Zustand stores backed by `persist` +
`createWorkspaceAwareStorage`, matching the pattern used by the
Issues view store. Selection now survives list → detail → back
navigation and page reloads, scoped per workspace.
Only `scope` is persisted; `search`, `sort`, and other ephemeral
filters intentionally still reset on remount.
Co-authored-by: multica-agent <github@multica.ai>
* fix(views): reset scope to mine when switching to a workspace with no persisted value
zustand persist.rehydrate() is a no-op when storage returns null, so
workspaces with no entry kept the previous workspace's in-memory scope
("all" leaked from one workspace into the next). Provide a custom merge
that resets to the default "mine" when no persisted state is present.
Add coverage for the missing-storage workspace-switch case for both
Agents and Squads.
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
* feat(settings): view/edit toggle for repositories tab
Saved repos render as static rows (truncated, monospace) with hover/focus-revealed
Edit + Delete affordances. Clicking Edit flips to the existing Input; on
successful Save the row returns to display mode. Save button is gated on a
dirty check (URL arrays in order) so a clean state reads as "All changes
saved". Resolves user feedback that the always-visible input made saved
state ambiguous (MUL-2217).
- Track editingIndices with a Set; new rows auto-enter edit mode; deleting
a row remaps indices so the wrong row never opens.
- Touch devices and focus-within keep the action buttons reachable.
- New i18n keys in en + zh-Hans (saved_hint, empty, edit/delete_aria, url_empty).
Co-authored-by: multica-agent <github@multica.ai>
* fix(settings): add Cancel affordance to exit clean edit mode
Clicking Edit on a clean saved row opened the row in edit mode with
no way back to display mode unless the user changed the URL and saved,
re-introducing the original saved-state ambiguity after an accidental
click. Add a per-row Cancel (X) button visible only in edit mode that:
- reverts the URL to the saved value for existing rows
- removes the row entirely for never-saved (newly added) rows
- exits edit mode without dirtying Save
Action group is always visible (no hover gate) while editing so the
exit is discoverable. Adds en/zh-Hans cancel_aria string and three
regression tests covering clean-cancel, dirty-cancel, and new-row-cancel.
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
- Add Squads to Features list (EN/zh) highlighting team-level agent routing
- Add a short Squads callout to the 'What is Multica?' section
- Remove the outdated 'Multica vs Paperclip' section from both READMEs
Co-authored-by: multica-agent <github@multica.ai>
* fix(daemon): resolve agent CLIs via login shell when daemon PATH misses them
GUI-launched daemons on macOS/Linux do not inherit the user's interactive
shell PATH, so fnm/nvm/volta multishells and the Anthropic native installer
silently disappear during onboarding even though `claude --version` works
in Terminal. Fall back to `$SHELL -ilc` to ask the login shell for the
canonical absolute path, then verify it with exec.LookPath before trusting
it. Symlinks (fnm/nvm prefix dirs) are resolved while the helper shell is
still alive so per-session paths get canonicalised before they vanish.
Refs MUL-2167, multica-ai/multica#2512.
Co-authored-by: multica-agent <github@multica.ai>
* fix(daemon): strip alias shadowing, harden timeout, lazy-resolve via login shell
Three follow-ups from the PR #2620 review (Elon):
1. Alias shadowing — `command -v claude` in zsh/bash returns the alias
definition, not the binary, and the absolute-path filter then rejects it.
The script now `unalias`/`unset -f` the name before lookup so `command -v`
falls through to the real PATH binary. This is the exact case behind
#2512.
2. Hard timeout — `CommandContext` kills only the shell process. Rc files
that background processes inheriting stdout (`direnv hook`, `nvm` shims,
plain `&`) keep the pipe open and `cmd.Output()` would block for as long
as the survivors live. `Cmd.WaitDelay` forcibly closes the pipes once
the cap elapses, so total startup penalty is bounded by
`timeout + waitDelay` regardless of rc-file content.
3. Lazy fallback — the resolver no longer runs on every daemon start.
`getShellResolved` is `sync.Once`-guarded and only fires when a bare
command name actually misses `exec.LookPath`. Users whose PATH already
contains every agent never pay the rc-file load cost.
Tests: - `TestResolveAgentsViaLoginShell_StripsAliasShadowing` — rc declares
`alias fakeclaude=...`, real binary lives on PATH, resolver must
return the binary, not the alias text.
- `TestResolveAgentsViaLoginShell_HardTimeoutOnBackgroundedStdout` —
rc backgrounds a 60s sleeper holding stdout; resolver must return
inside `timeout + waitDelay + slack`, not 60s.
- `TestLoadConfig_SkipsLoginShellWhenLookPathSucceeds` — when
exec.LookPath finds every agent, SHELL (a marker-writing sentinel)
must not be invoked.
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
* fix(issues): file-card render for self-host with local storage
Fixes#1520. When self-hosting without S3, the upload handler returns
site-relative URLs like /uploads/workspaces/<wsId>/<file>. Four
frontend regexes only matched https?://, so persisted
!file[name](/uploads/...) markdown failed to parse and leaked through
as raw text in the issue view, chat, skill file viewer, and board
card preview.
Narrow allow-list: the relative branch only accepts /uploads/ — not
any /-prefixed href — so protocol-relative //evil.com/x, path-traversal
/../api/x, and other internal /api/... paths are rejected. Without
this, a stored file-card with an attacker-chosen filename and a
//host/x href would turn into a one-click external-site jump via
window.open from inside an issue (per review feedback on #2349).
Single source of truth: packages/ui/markdown/file-cards.ts now exports
isAllowedFileCardHref + FILE_CARD_URL_PATTERN. The four sites use one
of them, so the next regression is cheaper than restoring four parallel
regexes.
- packages/ui/markdown/file-cards.ts: helper + URL pattern.
- packages/views/editor/extensions/file-card.tsx: Tiptap tokenizer
composes from FILE_CARD_URL_PATTERN.
- packages/views/editor/readonly-content.tsx: sanitiser uses helper.
- packages/ui/markdown/Markdown.tsx: sanitiser uses helper.
- packages/views/issues/components/board-card.tsx: strip markdown
tokens from the line-clamped board preview so raw !file[...] no
longer leaks there either.
- packages/ui/markdown/file-cards.test.ts: covers accept (/uploads/ok,
https://cdn/x) and reject (javascript:, data:, //evil.com/x,
/../api/x, /api/x, empty, ftp:, bare 'uploads/x') for both the
helper and the parser composed from the pattern.
javascript:, data:, and other dangerous schemes remain rejected.
* test(markdown): move file-card href allow-list test into @multica/views
Per review feedback on #2349: keep the test where vitest is already
running instead of bootstrapping a new test runner inside @multica/ui.
The test now lives at packages/views/editor/file-card-href.test.ts and
imports isAllowedFileCardHref / FILE_CARD_URL_PATTERN /
preprocessFileCards from the @multica/ui/markdown public surface,
exercising the same 30 cases.
Reverts the @multica/ui package.json test script + vitest devDep + the
local vitest.config.ts that the previous commit added; the package
goes back to typecheck + lint only, matching every other ui-only
package in the monorepo.
---------
Co-authored-by: Lalbadshah <11599756+Lalbadshah@users.noreply.github.com>
Rename the Deployment type dropdown options to Official App and
self-host so reporters pick the right one without guessing.
MUL-2212
Co-authored-by: multica-agent <github@multica.ai>
* refactor(agents): drop template chooser from create-agent dialog
Removes the blank-vs-template chooser, the template picker, and the
template detail step. The "Create agent" entry point now opens directly
on the form. The createAgentFromTemplate API and types remain
untouched — this only removes the UI entry.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Co-authored-by: multica-agent <github@multica.ai>
* docs(squads): fix stale comment about createAgentFromTemplate
Squad-scoped create flow no longer goes through the template path;
the dialog now only calls api.createAgent then api.addSquadMember.
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>
Co-authored-by: multica-agent <github@multica.ai>
Adds a dedicated bilingual /docs/squads page covering the squad model
(leader + members), assignment, comment trigger rules, archive
semantics, and the squad CLI surface. Wires the new page into
meta.json and meta.zh.json under the Agents section, and adds
short cross-references from agents, assigning-issues,
mentioning-agents, and the CLI reference so users can discover
squads from the pages they're already on.
MUL-2206
Co-authored-by: multica-agent <github@multica.ai>
When the user opens quick-create with a squad selected, the task is
enqueued against the squad's leader agent — but the squad, not the
leader, is the expected owner. The prompt previously instructed the
leader to "default to YOURSELF" using its own agent UUID, hiding new
issues from the squad's delegation flow.
Surface the squad's id + name on the claim response and branch the
default-assignee instruction in buildQuickCreatePrompt: when SquadID is
present, point --assignee-id at the squad UUID and explicitly forbid
self-assignment.
MUL-2203
Co-authored-by: multica-agent <github@multica.ai>
* feat(squads): add agent live peek hover card on member avatars
Squad members tab now opens a live-state peek card on agent avatar
hover/focus — workload, current issue (clickable), and last activity.
Identity (description / runtime / skills / owner) stays on the existing
AgentProfileCard; new AgentLivePeekCard is the second `hoverCardVariant`
on ActorAvatar so the 23+ existing profile-card call sites keep their
behaviour. Reuses the workspace agent-task snapshot already fetched by
the presence dot, so this adds zero new requests per row. Failed
terminal tasks surface as a small ⚠ on the last-activity line without
polluting workload (workload stays current-state only, matching the
deliberate split documented in core/agents/types.ts).
Co-authored-by: multica-agent <github@multica.ai>
* fix(squads): only enable hover card for agent avatars
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
Exposes the existing /api/tasks/{id}/cancel backend endpoint as a CLI
command. Combined with upstream #2107 (cancel running agent on
server-side task delete), this gives operators a way to interrupt a
runaway agent push-storm without resorting to admin-bypass on the
downstream PR.
Use cases:
- Titan / DevBot iterating beyond its boundary (e.g. push-skip loops)
- Codex turn that locked in tool-call spam
- Manual recovery when a long-running task needs to stop NOW
Symmetric with 'issue rerun': accepts the short ID prefix shown by
'issue runs', supports --issue scoping, and reuses resolveTaskRunID
for ambiguity handling.
Refs: PR#19 octo-server post-mortem (2026-05-13)
Co-authored-by: yujiawei <yujiawei@mininglamp.com>
* feat(squads): add tooltips and agent detail link to squad member row
Replace native title attributes on the make-leader and remove buttons
with proper Tooltip components, and add a new icon button on agent
rows that navigates to the agent detail page. All three tooltips are
localised.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Co-authored-by: multica-agent <github@multica.ai>
* fix(squads): keyboard focus visibility + AppLink for agent detail
- Add group-focus-within:opacity-100 so Tab to the row's hover-only
action buttons makes the container visible (previously opacity-0
kept buttons focusable but invisible).
- Replace the agent-detail jump button's onClick+push() with AppLink
href, restoring middle/Cmd+Click new-tab behavior. Removes the
now-unused onViewAgent callback chain.
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>
Co-authored-by: multica-agent <github@multica.ai>
Drop mx-auto + max-w-2xl wrappers around the Members and Instructions
tab content so the right pane fills the available width like the agent
detail page (TabContent uses flex h-full flex-col p-4 md:p-6).
Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>
Co-authored-by: multica-agent <github@multica.ai>
Text/code attachments (markdown, JSON, .ts, .log, …) need an attachment id
to render through `/api/attachments/{id}/content`. The composer pipeline
was dropping that id at the upload-hook boundary, so the Eye preview gate
only fired for media (PDF / video / audio via filename fallback).
- `useFileUpload` now returns the full `Attachment` (with `link` kept as a
`url` alias) so editor providers can resolve content-type and id.
- New-comment and reply composers hold a `pendingAttachments` state and
feed it to `ContentEditor`; the active subset (those still referenced in
the markdown) is sent on submit as before.
- Comment edit modes (CommentRow + CommentCardImpl) merge pending uploads
with `entry.attachments` for the editor and pipe `attachment_ids` into
`onEdit` so newly uploaded files actually bind to the comment.
- Issue description editor pushes pending `attachment_ids` on every
debounced save and invalidates `issueKeys.attachments` so the preview
Eye survives a refresh.
- `UpdateComment` and `UpdateIssue` handlers accept `attachment_ids` and
call the existing `linkAttachmentsByIDs` / `linkAttachmentsByIssueIDs`
helpers; the bind is idempotent so re-sending an existing id is safe.
Closes MUL-2153.
Co-authored-by: multica-agent <github@multica.ai>
* fix: trigger squad leader agent run when squad is @mentioned in comment
Previously, enqueueMentionedAgentTasks only processed m.Type == "agent"
mentions, skipping squad mentions entirely. The shouldEnqueueSquadLeaderOnComment
path only fires when the issue is already assigned to a squad.
This adds handling for m.Type == "squad" in enqueueMentionedAgentTasks:
when a squad is @mentioned, look up the squad's leader agent and enqueue
a task for them (with the same dedup/self-trigger/archived guards as
direct agent mentions).
Co-authored-by: multica-agent <github@multica.ai>
* fix: add canAccessPrivateAgent gate to squad mention branch
Closes the P1 permission vulnerability where a plain workspace member
could trigger a private squad leader by @mentioning the squad, bypassing
the private-agent access check that the direct @agent mention path
enforces.
Adds regression test TestCreateComment_SquadMentionPrivateLeaderBlocksPlainMember.
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
* feat(agents): rewrite template catalog as 25 lightweight starters
Replaces every Phase-1 template with a curated set built around the
"persona + intake + scaffold + hard negatives" instruction shape. Cross-
platform survey (Cursor / Cline / Roo / Continue / Custom GPTs) showed
the industry baseline for starter agents is "few but sharp" — single
intent, no methodology buy-in, mostly prompt-only. The original catalog
went the opposite direction (avg 2.5 skills, six-skill Full-stack
methodology stack) and felt heavy for first-time use.
Catalog shape:
- 25 templates across 7 categories: Engineering (8), Product (4),
Writing (5), Design (3), Communication (2), Team (1), Productivity (2).
New Product / Design / Communication / Team domains fill gaps the old
Eng-heavy catalog ignored.
- 16 / 25 are prompt-only (no skill fan-out). Avg 0.56 skill per template
vs. 2.5 prior. Heaviest is 2 skills, only for templates whose intent
cannot be expressed in instructions alone (Playwright runner, single-
file HTML bundlers, design + UX-guidelines pair).
- Universal top-frequency intents that the old catalog missed are now
covered: Code Explainer (intent #1 across every platform surveyed),
Translator (中英), Summarizer, Writing Critic, PRD Drafter/Critic,
RCA Writer, ADR Writer, PR Description Writer, Commit Message Writer.
Loader allows 0-skill templates:
- server/internal/agenttmpl/loader.go drops the "must declare at least
one skill" validation; comment explains the picker's "Prompt only"
rendering path.
- loader_test.go: removed the corresponding negative case, added
TestLoadFromFS_PromptOnlyTemplate as a regression guard.
- agent_template.go handler is unchanged — every len(tmpl.Skills) call
site was already 0-safe (empty fan-out short-circuits the fetch phase
and the in-tx loop both skip cleanly).
Frontend:
- template-picker.tsx: 18 new lucide icons (BookOpen, Bug, GitPullRequest,
GitCommit, AlertTriangle, Scale, ClipboardList, Microscope, UserRound,
Target, Highlighter, Languages, AlignLeft, GraduationCap, Lightbulb,
Type, MessageSquare, Briefcase). Card renders a "Prompt only" badge
when skills.length === 0 instead of "0 skills".
- template-detail.tsx: skill list section is hidden entirely for prompt-
only templates — a header reading "Includes 0 skills" above an empty
list was just visual noise. Instructions section below carries the
agent's identity for these.
- locales/en + zh-Hans agents.json: new create_dialog.template_card.
prompt_only key ("Prompt only" / "纯指令").
Verification:
- go test ./internal/agenttmpl/ — 9/9 pass, including
TestLoad_RealTemplates which fails closed if any new JSON is malformed.
- pnpm typecheck — all 6 packages clean.
- pnpm --filter @multica/views test — 482/482 pass.
- pnpm lint — 0 errors.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* feat(agents): add category filter pills to template picker
25 templates across 7 categories made the picker scroll-heavy on first
open. Add a single-select category filter row above the grid so a PM
can isolate Product templates in one click, an engineer can jump
straight to Engineering, etc.
Visual reuses the IssuesHeader scope-toggle pattern verbatim — Button
variant="outline" + active class swap (bg-accent / text-muted-foreground)
— so the affordance reads the same as the existing filter pills in
issues / squads / runtimes / my-issues. flex-wrap keeps the 8 pills
(All + 7 categories) honest on narrow widths.
Counts are inlined into the label ("Engineering (8)") rather than
shown as a separate badge — single-line-tall pills look right next to
the picker grid, and surfacing the per-category density up front
doubles as a hint at the catalog's "less but sharper" intent.
When a specific category is active, the grid renders flat (no
section headers) — the active pill already names what's on screen,
and a header reading "Engineering" above an only-Engineering grid is
visual duplication. "All" falls back to the prior grouped layout.
State is component-local (no URL sync, no persistence) since the
picker is dialog-internal transient state — closing the dialog
naturally resets the filter, which is the expected behaviour for a
"choose from a catalog" surface.
i18n: new `create_dialog.template_picker.filter_all` key in en + zh.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Adds a Create Agent button on the Squad detail Members tab, visible
only to workspace owner/admin (matching the AddSquadMember backend
gate). The dialog reuses the existing CreateAgentDialog — both the
manual and template paths now accept an optional squadId; when set,
the dialog runs addSquadMember after createAgent / createAgentFromTemplate
and skips the navigation to the agent detail page so the user lands
back on the Members tab.
Atomicity is best-effort frontend-serial (no new backend transaction):
on partial failure the dialog surfaces a warning toast and the agent
remains addable from the existing Add Member flow.
Co-authored-by: multica-agent <github@multica.ai>
* fix: execution log name rendering and squad assignee support
- Strip mention markdown in trigger_summary ([@Name](mention://...) → @Name)
so execution log rows show clean text instead of raw markdown
- Add squad to ActorFilterValue type so squad assignees are filterable
- Add squad section to assignee filter dropdown in issues-header
- Add i18n keys for squads_group (en/zh-Hans)
Co-authored-by: multica-agent <github@multica.ai>
* fix: address PR #2575 review feedback
1. Extract stripMentionMarkdown as reusable helper with proper regex
- Handles escaped brackets in names (e.g. David\[TF\])
- Skips backslash-escaped mentions (\[@...])
- Handles issue mentions (no @ prefix)
- Does not touch regular markdown links
- 10 unit tests added
2. Squad only appears in Assignee filter, not Creator
- Added showSquads prop to ActorSubContent (default true)
- Creator filter passes showSquads={false}
3. Squad included in Agents scope
- issues-page scope filter now includes squad in agents scope
- 2 regression tests added for scope coverage
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
The per-turn prompt in buildCommentPrompt() only injected the squad
leader no_action prohibition inside the 'if TriggerAuthorType == agent'
block. When a member (human) posted a comment like 'LGTM', the squad
leader was triggered but the per-turn prompt did NOT include the
prohibition, causing the model to post noise comments like 'LGTM is a
pure acknowledgment — no reply needed. Exiting silently.'
Fix: move the squad leader no_action rule outside the agent-only block
so it fires for ALL trigger types (agent and member).
Fixes: MUL-2168
Co-authored-by: multica-agent <github@multica.ai>
* feat: support pinyin search in @mention suggestions
Add pinyin matching for Chinese names in the mention suggestion popup.
Users can now search by:
- Full pinyin: 'liyunlong' matches '李云龙'
- Initial letters: 'lyl' matches '李云龙'
- Partial/hybrid: 'liyu' or 'liyunl' matches '李云龙'
Implementation:
- New pinyin-match.ts utility using pinyin-pro library
- Integrated into member, agent, and squad filters in mention-suggestion.tsx
- 21 tests passing (9 unit + 12 integration)
Co-authored-by: multica-agent <github@multica.ai>
* fix: normalize ü→v in pinyin matching for names like 吕布
Enable pinyin-pro's v:true option so 吕→lv instead of lü.
Add test case for 吕布/lvbu matching.
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
PR #2564 only added IsSquadLeader handling to the assignment-triggered
workflow path and the Output section. When a squad leader is triggered by
a comment (the common case for re-evaluation), the comment-triggered
workflow path had NO squad leader special handling, so the model still
posted comments announcing no_action/silence.
Changes:
- runtime_config.go: Add IsSquadLeader check to comment-triggered step 4
with explicit prohibition against posting no_action announcement comments
- runtime_config.go: Strengthen Output section from 'may exit silently' to
'MUST exit without posting any comment' with explicit DO NOT examples
- runtime_config.go: Strengthen assignment-triggered step 5 similarly
- prompt.go: Add squad leader no_action rule to per-turn comment prompt
when trigger author is an agent and agent instructions contain the
Squad Operating Protocol marker
- Add tests for both the per-turn prompt and CLAUDE.md generation
Fixes MUL-2168
Co-authored-by: multica-agent <github@multica.ai>
* fix(squad): skip leader on comment when a member @mentions any agent (MUL-2170)
When a human commenter routes an issue directly at a specific agent via
[@Name](mention://agent/<id>), the squad leader was still being woken up
to evaluate the same comment. The leader's only real options were to
re-delegate to the agent the member already named or to record
no_action — both of which produce queue noise without changing the
outcome.
This skips the leader-enqueue path entirely when:
- the assignee is a squad,
- the comment author is a member, AND
- the comment body contains at least one agent mention.
Agent-authored comments are intentionally exempt: when an agent posts
an update that @mentions another agent, the leader still needs to
coordinate the thread. The existing leader-self-trigger guard is
preserved. Only the current comment's body is inspected — parent
(thread root) mentions are not inherited here.
Tests cover the helper (mentions parsing) plus the integration matrix:
member plain / member @member / member @non-leader-agent /
member @leader / agent @agent / leader-self.
Co-authored-by: multica-agent <github@multica.ai>
* test(squad): exercise full CreateComment path for leader-skip rule (MUL-2170)
Adds an integration test that drives the HTTP-layer CreateComment handler
(not just the helper) to lock the call-site wiring: a member top-level
comment with an @agent skips the squad leader, and a subsequent plain
reply in the same thread DOES wake the leader — the parent's @agent
mention must not be inherited into the leader-skip decision.
Picks up a non-blocking review note on PR #2569.
Co-authored-by: multica-agent <github@multica.ai>
* fix(squad): skip leader on any explicit member mention, not only @agent (MUL-2170)
Broaden the leader-skip rule for squad-assigned issues: a member comment
that explicitly @mentions anyone — @agent, @member, @squad, or @all —
counts as deliberate routing and the squad leader stays out. Issue
cross-references (mention://issue/...) are not routing and still trigger
the leader as before.
Per Bohan's follow-up on MUL-2170 — @member should suppress the leader
for the same reason @agent does: the human has already pointed at a
specific recipient, so a leader turn would just be observation noise.
Helper renamed commentMentionsAnyAgent → commentMentionsAnyone with
explicit handling of all four routing mention types. Existing call-site
wiring (current-comment-only, agent-author exemption, leader self-trigger
guard) is unchanged.
Tests updated and extended to cover the full routing matrix:
@member / @squad / @all / @issue (cross-ref) plus the @agent variants
already covered.
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
The Eye button required a fully resolved Attachment record (URL-lookup
via `resolveAttachment(href)`) before showing. Download only required
the URL, falling back to `openExternal(href)` when the lookup missed.
Result: any case where the URL in markdown couldn't be reverse-matched
to the entity's `attachments` prop (cross-comment copy-paste, stale
caches) silently hid the Preview button while Download kept working —
edit and readonly surfaces diverged for the same content.
Widen the Preview gate to mirror Download: show the Eye whenever the
filename indicates a previewable type. Introduce a `PreviewSource`
tagged union — `{ kind: "full", attachment }` for the existing path,
`{ kind: "url", url, filename }` for the fallback. Media kinds
(pdf/video/audio) render directly from the URL; text kinds still
require an attachment id because the /content proxy is ID-keyed, so
`tryOpen` rejects URL+text combinations and PreviewContent has a
defensive fallback for direct mounts.
Side effects:
- `getPreviewKind` gains filename-extension fallbacks for video/audio
(was PDF-only); without these the URL-only path can't infer kind
when content_type is empty.
- AttachmentList in comment-card.tsx unchanged behaviorally — only the
tryOpen call site is updated to the new signature.
Pre-existing architectural issues (AttachmentList readonly-only,
URL-based attachment lookup, per-entity ownership) are intentionally
out of scope.
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Template create used to silently default the runtime to "first usable"
and never collected a model — users had no idea where the new agent
would run or which model it would use until they opened the detail
page. Add a Runtime + Model picker pair above the skill list on the
template-detail step so the choice is visible (and overridable) before
the one-click Use action.
- Extract RuntimePicker out of create-agent-dialog so the form and the
template-detail step share one popover; selection seeding moves into
the picker too, since it's the only place that knows the active
filter (mine/all). Parent keeps just the duplicate-mode pre-fill.
- Mirror RuntimePicker's label-row + trigger DOM in ModelDropdown so
the two pickers render at identical heights when sat side-by-side
(fixes a 6-8px misalignment caused by inconsistent label-row sizing).
- Send model in createAgentFromTemplate; server side already accepts
the field (CreateAgentFromTemplateRequest.Model, omitempty), empty
string still falls through to the runtime's default model.
- Drop the runtime_register_first fallback hint that made the Runtime
trigger two-line in the empty state, breaking alignment with Model's
one-line trigger.
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The runtime prompt's Output section unconditionally required all tasks to
post a comment via 'multica issue comment add', which conflicted with the
squad leader protocol that says to 'exit silently' on no_action.
Changes:
- Add IsSquadLeader bool to TaskContextForEnv (detected via Squad Operating
Protocol marker in agent instructions)
- Relax the Output section and assignment-triggered workflow step 5 to
allow squad leaders to exit with only a 'multica squad activity' call
when the outcome is no_action
Fixes MUL-2168
Co-authored-by: multica-agent <github@multica.ai>
* fix(cli): resolve squad assignees in issue create/update/assign (MUL-2165)
The CLI assignee resolver only searched workspace members and agents, so a
quick-create input like "assign to <SquadName>" silently fell through to
"Unrecognized assignee: <SquadName>" in the issue description — even though
squads are first-class assignees server-side and the prompt's whole point was
to route the work for the user.
Extend resolveAssignee / resolveAssigneeByID to also fetch /api/squads, teach
the actor display lookup to render squad names in table output, update the
quick-create prompt and runtime-config command listing to mention
`multica squad list` alongside members and agents, and lock in the new
behavior with tests.
Co-authored-by: multica-agent <github@multica.ai>
* fix(cli): gate squad assignee resolution behind an allowed-kinds set (MUL-2165)
The earlier MUL-2165 fix taught resolveAssignee / resolveAssigneeByID to also
return (squad, ...), but those helpers are shared. Project lead and issue
subscriber callers were still using them, and their target schemas reject
squads — project.lead_type has a DB CHECK constraint
(server/migrations/034_projects.up.sql:10) and the subscriber handler's
isWorkspaceEntity switch only knows member/agent
(server/internal/handler/handler.go:414). So
`multica project create --lead "<SquadName>"` and
`multica issue subscriber add --user "<SquadName>"` would resolve to
(squad, ...) and surface as a 500/403 server-side instead of a clean
CLI-side resolution error.
Thread an assigneeKinds set through the resolver and the pickAssigneeFromFlags
helper. Issue create/update/assign/list pass `issueAssigneeKinds` (all three);
project lead and subscriber pass `memberOrAgentKinds`. The squads fetch is
skipped entirely when not allowed, and the not-found / no-match error wording
adapts to the allowed kinds so it never mentions a type the caller cannot use.
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
* feat(quick-create): searchable actor picker + squad support (MUL-2163)
- Replaces the flat agent dropdown in the "Create with agent" modal with a
searchable PropertyPicker that lists Agents and Squads in separate
sections, so users can filter by name and pick a squad as the creator.
- Persists the selection as (lastActorType, lastActorId), removing the
agent-only lastAgentId field on the quick-create store.
- Adds squad_id to the quick-create API request and stamps it onto the
task's QuickCreateContext. The handler resolves the squad to its leader
agent (re-using validateAssigneePair) and the daemon claim path injects
the squad-leader briefing when the task carries a squad hint, matching
the behavior of issue-bound squad tasks.
Co-authored-by: multica-agent <github@multica.ai>
* fix(create-issue): forward squad picks across manual→agent switch
Manual mode → agent mode previously only carried `agent_id`, so picking
a squad and then flipping to agent silently fell back to the persisted
actor / first visible agent and lost the user's choice. Carry `squad_id`
on the same branch so the agent panel honors the squad pick.
Adds a sibling test alongside the existing project-carry case.
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
* fix(issues): unify assignee menu with shared AssigneePicker (MUL-2157)
The Assignee submenu inside IssueActionsMenuItems was a parallel
implementation: no search, no squads, no agent permission check, no
archive filter, no frequency sort. The divergence was most visible from
the Inbox (where the issue detail's sidebar starts collapsed, so users
reach for the 3-dot menu).
Replace the submenu with a single menu item that closes the
surrounding dropdown / context menu and hands off to the shared
AssigneePicker popover — same component already used in the issue
detail sidebar, board cards, batch toolbar, and create-issue modal.
The picker is conditionally mounted to avoid every row in list / board
views subscribing to the members / agents / squads / frequency queries
on mount.
Co-authored-by: multica-agent <github@multica.ai>
* test(issues): mock squadListOptions + add Assignee picker handoff test
`AssigneePicker` reads `squadListOptions` and `assigneeFrequencyOptions`
from `@multica/core/workspace/queries`. Tests that render IssueDetail
or IssueActionsDropdown without those mocks throw at the picker's
useQuery call and cascade into unrelated assertion failures — this is
what was leaving the `@multica/views` test job red on the MUL-2157 PR.
Add the missing mocks. Add a regression test that clicks the Assignee
menu item and asserts the shared picker (search input + Members group)
takes over, so a future regression to the parallel-implementation bug
this PR fixes fails loudly instead of silently.
Co-authored-by: multica-agent <github@multica.ai>
---------
Co-authored-by: multica-agent <github@multica.ai>
* feat(usage): mirror Tokens metric toggle onto Usage page Daily chart (MUL-2148)
#2537 added the Cost/Tokens metric toggle to the Daily chart inside the
runtime-detail Usage section (packages/views/runtimes/components/
usage-section.tsx). The workspace-level Usage page at /{slug}/usage
imports the same DailyCostChart primitive but renders it from
dashboard-page.tsx without any toggle wrapper, so #2537 only landed on
half of the surface that says "Daily cost".
This PR mirrors the same pattern to dashboard-page.tsx so users see
the toggle wherever a "Daily" chart appears.
Changes
- `packages/views/dashboard/utils.ts`: new `aggregateDailyTokens` helper
that folds DashboardUsageDaily[] into the same DailyTokenData[] shape
the DailyTokensChart consumes (mirrors aggregateByDate's dailyTokens
branch from the runtimes side, adapted to DashboardUsageDaily field
names).
- `packages/views/dashboard/components/dashboard-page.tsx`: rename
`DailyCostBlock` → `DailyTrendBlock`, add a Cost/Tokens Segmented
next to the section title, switch chart and title based on the
active metric, per-metric empty-state (so a workspace with unmapped
pricing but recorded tokens still gets a real Tokens chart while
the Cost view falls through to the empty-state — same convention as
DailyTab in usage-section.tsx).
- usage.json (en + zh-Hans): split `daily.title` into `title_cost` +
`title_tokens`, add `metric_cost` + `metric_tokens` toggle labels.
* feat(usage): default Daily chart to Tokens metric
Most users land on /{slug}/usage to gauge "how much agent work
happened" rather than "how much was spent." Tokens is the more
universally meaningful axis on first read (Cost depends on having
pricing mapped for every model and on whether the workspace has
unmaintained models). Cost stays one click away via the same toggle.
Also reorder the Segmented so Tokens sits first, matching the new
default.
2026-05-13 22:07:47 +08:00
455 changed files with 36457 additions and 4776 deletions
@@ -32,6 +32,8 @@ Multica turns coding agents into real teammates. Assign issues to an agent like
No more copy-pasting prompts. No more babysitting runs. Your agents show up on the board, participate in conversations, and compound reusable skills over time. Think of it as open-source infrastructure for managed agents — vendor-neutral, self-hosted, and designed for human + AI teams. Works with **Claude Code**, **Codex**, **GitHub Copilot CLI**, **OpenClaw**, **OpenCode**, **Hermes**, **Gemini**, **Pi**, **Cursor Agent**, **Kimi**, and **Kiro CLI**.
For larger teams, Squads add a stable routing layer: assign work to a group led by an agent, and the leader delegates to the right member.
@@ -53,6 +55,7 @@ Like Multics before it, the bet is on multiplexing: a small team shouldn't feel
Multica manages the full agent lifecycle: from task assignment to execution monitoring to skill reuse.
- **Agents as Teammates** — assign to an agent like you'd assign to a colleague. They have profiles, show up on the board, post comments, create issues, and report blockers proactively.
- **Squads** — group agents (and humans) under a leader agent and assign work to the *squad*. The leader decides who should pick it up, so routing stays stable as the team grows. `@FrontendTeam` instead of `@alice-or-bob-or-carol`.
- **Autonomous Execution** — set it and forget it. Full task lifecycle management (enqueue, claim, start, complete/fail) with real-time progress streaming via WebSocket.
- **Reusable Skills** — every solution becomes a reusable skill for the whole team. Deployments, migrations, code reviews — skills compound your team's capabilities over time.
- **Unified Runtimes** — one dashboard for all your compute. Local daemons and cloud runtimes, auto-detection of available CLIs, real-time monitoring.
@@ -128,21 +131,6 @@ Create an issue from the board (or via `multica issue create`), then assign it t
---
## Multica vs Paperclip
| | Multica | Paperclip |
|---|---------|-----------|
| **Focus** | Team AI agent collaboration platform | Solo AI agent company simulator |
| **User model** | Multi-user teams with roles & permissions | Single board operator |
@@ -25,14 +25,30 @@ These have sensible defaults and only need to be set when tuning a large or cons
### Email (Required for Authentication)
Multica uses email-based magic link authentication via [Resend](https://resend.com).
Multica supports two emailbackends. `SMTP_HOST` takes priority when set; otherwise `RESEND_API_KEY` is used. With neither configured, verification codes are printed to the server log — copy them from there to log in.
#### Option A: Resend (recommended for cloud deployments)
> **Note:** If Resend is not configured, generated verification codes are printed to backend logs. A fixed local testing code is disabled by default; to opt in on a private test instance, set `APP_ENV=development` and `MULTICA_DEV_VERIFICATION_CODE` to a 6-digit value. It is ignored when `APP_ENV=production`.
Use this option when your deployment cannot reach the public internet or you already have an internal mail relay (e.g. Exchange, Postfix, SendGrid on-prem).
| `SMTP_TLS_INSECURE` | Set `true` to skip TLS certificate verification (self-signed / private CA certs) | `false` |
STARTTLS is used automatically when advertised by the server. Port 465 (SMTPS / implicit TLS) is not currently supported - use ports 25 or 587 with STARTTLS.
> **Note:** If neither Resend nor SMTP is configured, generated verification codes are printed to backend logs — copy them from there to log in. A fixed local testing code (e.g. `888888`) is **opt-in only**: set `MULTICA_DEV_VERIFICATION_CODE=888888` in `.env` and keep `APP_ENV` non-production. The Docker self-host stack pins `APP_ENV=production`, so the shortcut is ignored there. **Never enable a fixed code on a publicly reachable instance.**
@@ -5,7 +5,7 @@ description: Hand an issue to an agent and it takes over as the official assigne
import { Callout } from "fumadocs-ui/components/callout";
Assign an [issue](/issues) to an [agent](/agents) and it works as the **official assignee** until the work is done — it can read the full issue context (description + all [comments](/comments)) and change status, post comments, and edit fields. This is the **most common and heaviest** of Multica's four trigger paths.
Assign an [issue](/issues) to an [agent](/agents) and it works as the **official assignee** until the work is done — it can read the full issue context (description + all [comments](/comments)) and change status, post comments, and edit fields. This is the **most common and heaviest** of Multica's four trigger paths. The same flow also accepts a [squad](/squads) as the assignee — Multica then triggers the squad's **leader agent** instead.
| Path | When to use | Changes the issue | Context | Priority | Auto retry |
|---|---|---|---|---|---|
@@ -18,7 +18,7 @@ Assign an [issue](/issues) to an [agent](/agents) and it works as the **official
## Assign from the UI
On the issue detail page, click the **Assignee** picker. It lists every member in the workspace plus all non-archived agents. Pick an agent and the issue is assigned right away.
On the issue detail page, click the **Assignee** picker. It lists every member in the workspace, all non-archived agents, and every non-archived [squad](/squads). Pick an agent (or squad) and the issue is assigned right away.
A few rules:
@@ -78,5 +78,6 @@ But **different agents can work on the same issue in parallel** — for example,
## Next
- [**@-mention an agent in a comment**](/mentioning-agents) — a lighter trigger that leaves assignee and status untouched
- [**Squads**](/squads) — assign to a group of agents and let the leader decide who picks it up
- [**Chat**](/chat) — one-to-one conversation outside any issue
- [**Autopilots**](/autopilots) — let agents start work automatically on a schedule
@@ -12,9 +12,11 @@ For the list of environment variables referenced below, see [Environment variabl
## How email + verification code sign-in works
The user enters an email on the sign-in page → the server sends a 6-digit code → the user enters it → the server verifies it → a JWT cookie is issued. Standard flow. It requires [Resend](https://resend.com/) as the email provider:
The user enters an email on the sign-in page → the server sends a 6-digit code → the user enters it → the server verifies it → a JWT cookie is issued. Standard flow. Two delivery backends are supported — pick whichever fits your deployment:
1. Create a Resend account and verify your domain
### Option A: Resend (recommended for cloud / public-internet deployments)
1. Create a [Resend](https://resend.com/) account and verify your domain
2. Create an API key
3. Set the environment variables:
@@ -25,7 +27,22 @@ The user enters an email on the sign-in page → the server sends a 6-digit code
4. Restart the server
**What happens if you don't set `RESEND_API_KEY`**: the server doesn't error, but **every email that should have been sent is written to the server's stdout only**. Handy for local development (copy the code from the logs); in production it's a black hole.
Use this when the deployment can't reach `api.resend.com` or you already have an internal mail relay (Exchange, Postfix, on-prem SendGrid, etc.). `SMTP_HOST` takes priority over `RESEND_API_KEY` when both are set.
```bash
SMTP_HOST=smtp.internal.example.com
SMTP_PORT=587 # default 25; use 587 for STARTTLS submission
SMTP_USERNAME=multica # leave empty for unauthenticated relay
SMTP_PASSWORD=...
SMTP_TLS_INSECURE=false # set true only for self-signed / private CA
RESEND_FROM_EMAIL=noreply@yourdomain.com # reused as the From: header
```
STARTTLS is upgraded automatically when the server advertises it. Port 465 (SMTPS / implicit TLS) is **not** currently supported — use port 25 or 587.
**What happens if you set neither**: the server doesn't error, but **every email that should have been sent is written to the server's stdout only**. Handy for local development (copy the code from the logs); in production it's a black hole.
## Fixed local testing codes
@@ -34,7 +51,7 @@ The user enters an email on the sign-in page → the server sends a 6-digit code
The old behavior where non-production instances accepted `888888` by default has been removed. Unless you explicitly configure it, typing `888888` is treated like any other wrong code.
Local development without Resend should use the generated code printed in server logs. If you need deterministic local/private automation, set `MULTICA_DEV_VERIFICATION_CODE` to a 6-digit value such as `888888`, and keep `APP_ENV` non-production:
Local development without any email backend configured (no Resend, no SMTP) should use the generated code printed in server logs. If you need deterministic local/private automation, set `MULTICA_DEV_VERIFICATION_CODE` to a 6-digit value such as `888888`, and keep `APP_ENV` non-production:
description: Let agents start work on a cron schedule — or trigger once manually via the UI or CLI.
description: Let agents start work on a cron schedule, an inbound webhook, or trigger once manually via the UI or CLI.
---
import { Callout } from "fumadocs-ui/components/callout";
@@ -16,13 +16,13 @@ Create a new autopilot on the workspace's **Autopilot** page. You set:
- **Priority** — inherited by the `task` it produces (same semantics as issue priority)
- **Description / prompt** — the work description the agent receives each run
- **Execution mode** — see below
- **Triggers** — at least one `schedule` (cron + timezone)
- **Triggers** — at least one `schedule` (cron + timezone) or `webhook`
## Pick an execution mode
An autopilot has two execution modes. **Start with "create issue" mode.**
- **Create issue mode** (`create_issue`) — default, **recommended**. Each trigger first creates an issue in the workspace (the title supports interpolation like `{{date}}`), then assigns the issue to the agent through the normal assignment flow. All work lands on the issue board with the same history, comments, and status as a manually assigned issue.
- **Create issue mode** (`create_issue`) — default, **recommended**. Each trigger first creates an issue in the workspace (the title currently supports a single placeholder, `{{date}}`, which interpolates to the UTC date in `YYYY-MM-DD` format; any other `{{...}}` token is rejected at create-time so a typo cannot silently land as the literal string in your issue titles), then assigns the issue to the agent through the normal assignment flow. All work lands on the issue board with the same history, comments, and status as a manually assigned issue.
- **Run-only mode** (`run_only`) — skips issue creation and enqueues a `task` directly. The run is invisible on the board — you can only see it in the autopilot's run history.
- The linked issue (create issue mode) or `task` (run-only mode)
- Failure reason (if failed)
- Failure reason (if failed or skipped)
## What happens when an autopilot fails
@@ -72,7 +166,11 @@ Why no auto-retry: autopilots are already periodic, so adding system-level retri
## What's not yet available
**Webhook and API triggers are not available yet.** The autopilot trigger schema reserves `webhook` and `api` types, but **they are not wired up to any ingress route** — the UI can create triggers of either type, but they will not actually fire. Today, **only `schedule` and manual triggers are end-to-end usable.**
**API-kind triggers are not wired up.** The trigger schema reserves an `api`
kind, but no ingress route fires it; the UI shows a Deprecated badge for
existing rows and offers no copy/rotate affordances. Per-trigger HMAC
signature verification, IP allowlists, and provider-specific event presets
are tracked as follow-ups; v1 URLs are bearer-only.
@@ -35,14 +35,28 @@ These are the core variables you must think about before deploying — some have
## Email configuration
Multica uses [Resend](https://resend.com/) to send verification codes and invite emails.
Multica supports two delivery backends — [Resend](https://resend.com/) for cloud deployments, or an SMTP relay for internal / on-premise networks. `SMTP_HOST` takes priority over `RESEND_API_KEY` when both are set.
### Resend
| Variable | Default | Description |
|---|---|---|
| `RESEND_API_KEY` | empty | Resend API key |
| `RESEND_FROM_EMAIL` | `noreply@multica.ai` | Sender address (must be a domain verified in your Resend account) |
| `RESEND_FROM_EMAIL` | `noreply@multica.ai` | Sender address (must be a domain verified in your Resend account; also reused as the `From:` header when SMTP is in use) |
**Behavior when `RESEND_API_KEY` is unset**: the server does not error, but every email that should have been sent (verification codes, invite links) **is written to the server's stdout only**. Convenient for local development — copy the code out of the server logs; **in production, forgetting to set this creates a silent black hole**, with users never receiving email and no error surfaced.
### SMTP relay
| Variable | Default | Description |
|---|---|---|
| `SMTP_HOST` | empty | SMTP relay hostname. Setting this activates SMTP mode and overrides Resend |
| `SMTP_PORT` | `25` | SMTP port. Use `587` for STARTTLS submission; **port 465 (SMTPS / implicit TLS) is not supported** |
| `SMTP_TLS_INSECURE` | `false` | Set `true` to skip TLS certificate verification (private CA / self-signed only) |
STARTTLS is upgraded automatically when the server advertises it. The dial timeout is 10s and the whole SMTP session has a 30s deadline, so a black-holed relay can't hang the auth handler.
**Behavior when neither is set**: the server does not error, but every email that should have been sent (verification codes, invite links) **is written to the server's stdout only**. Convenient for local development — copy the code out of the server logs; **in production, forgetting to set this creates a silent black hole**, with users never receiving email and no error surfaced.
## Google OAuth configuration
@@ -114,6 +128,25 @@ Three allowlist layers combine by priority. **If any layer is set to a non-empty
**Invite flows themselves do not check the signup allowlist** — but the invitee must still be able to **sign in** before accepting the invite. If they already have a Multica account (for example from another workspace), they can accept directly, unaffected by the allowlist; **if they have never signed up**, the first step of sign-in (requesting a verification code) still passes through the allowlist check, and an email rejected by `ALLOW_SIGNUP=false` or by `ALLOWED_EMAILS` / `ALLOWED_EMAIL_DOMAINS` **cannot finish signup, and therefore cannot accept the invite**.
## Rate limiting (optional Redis)
Public auth endpoints — `/auth/send-code`, `/auth/verify-code`, `/auth/google` — have per-IP fixed-window rate limiting in front of them. The limiter is backed by Redis. When `REDIS_URL` is unset the middleware is a **no-op** (fail-open) and the backend logs `rate limiting disabled: REDIS_URL not configured` at startup.
| Variable | Default | Description |
|---|---|---|
| `REDIS_URL` | empty | Redis connection URL (for example `redis://localhost:6379/0`). When unset, rate limiting on auth endpoints is disabled. The same Redis is also used by the realtime hub fan-out, the PAT cache, and the daemon-token cache — they all fall back to in-memory / direct-DB mode when unset |
| `RATE_LIMIT_AUTH` | `5` | Max requests per IP per minute against `/auth/send-code` and `/auth/google` |
| `RATE_LIMIT_AUTH_VERIFY` | `20` | Max requests per IP per minute against `/auth/verify-code` |
| `RATE_LIMIT_TRUSTED_PROXIES` | empty | Comma-separated CIDRs whose `X-Forwarded-For` header the limiter is allowed to trust. Empty (the default) means **never trust XFF** — the limiter only uses the direct connection's `RemoteAddr` |
When a request is over the limit, the server replies with `429 Too Many Requests`, `Retry-After: 60`, and body `{"error":"too many requests"}`.
<Callout type="warning">
**Behind a reverse proxy you must set `RATE_LIMIT_TRUSTED_PROXIES`.** Otherwise every real user shares the proxy's IP from the backend's point of view, the whole deployment ends up in one bucket, and `/auth/send-code` becomes 5 req/min for the entire site. Typical values: `127.0.0.1/32,::1/128` for a same-host Caddy / Nginx; the CDN's published ranges for Cloudflare / ALB / CloudFront. Only IPs whose `RemoteAddr` falls inside one of these CIDRs may use `X-Forwarded-For` to identify the client.
</Callout>
This separate `RATE_LIMIT_TRUSTED_PROXIES` is **not** the same as `MULTICA_TRUSTED_PROXIES`, which controls the autopilot-webhook limiter (`/api/webhooks/autopilots/{token}`). Each limiter parses its own list, so a deployment behind a proxy should set both.
## Daemon tuning parameters
The daemon runs on the user's local machine, and its config is read from local environment variables too. The common ones:
@@ -16,6 +16,10 @@ Same as mentioning a member — type `@` to open the picker and select an agent.
The `@mention` Markdown syntax, the picker, and `@all` semantics are covered in [**Comments**](/comments).
<Callout type="info">
**You can also `@`-mention a [squad](/squads) in a comment.** The same picker surfaces squads alongside members and agents; selecting one inserts `[@SquadName](mention://squad/<uuid>)` and triggers the squad's **leader agent** to coordinate a response — assignee and status stay untouched.
</Callout>
## How it differs from assignment
Both put the agent to work, but the mechanics are entirely different:
@@ -53,6 +57,7 @@ This guard **only blocks direct self-references.** Agent A @-mentioning agent B
## Next
- [**Squads**](/squads) — `@`-mention a squad to have the leader route the question to the right member
- [**Chat**](/chat) — one-to-one conversation outside any issue
- [**Autopilots**](/autopilots) — let agents start work automatically on a schedule
- [**Comments**](/comments) — `@mention` syntax, the picker, and `@all` semantics
**Ports listen on `127.0.0.1` only.** `docker-compose.selfhost.yml` binds every published port to loopback — `ss -tlnp` will not show `0.0.0.0:8080`, and the services are unreachable from other machines by design. The default `JWT_SECRET` and Postgres credentials must never sit on the open internet. For cross-machine access, front the stack with a reverse proxy that terminates TLS — see [Step 5b — Cross-machine: front with a reverse proxy](#5b-cross-machine-front-with-a-reverse-proxy).
</Callout>
## 2. Important: keep production safety on
<Callout type="warning">
@@ -59,7 +63,9 @@ Before any public deployment, make sure `.env` has `APP_ENV=production` and `MUL
Without email configured, your users can't receive verification codes by email; the server prints generated codes to stdout instead.
To actually send verification emails:
Two delivery backends are supported — pick whicheverfits your network:
**Option A — Resend (cloud / public-internet deployments):**
1. Sign up at [Resend](https://resend.com/) and get an API key
2. Verify a sending domain you control
@@ -70,36 +76,80 @@ To actually send verification emails:
**Option B — SMTP relay (internal networks / on-premise):**
For more auth configuration (OAuth, signup allowlist), see [Auth setup](/auth-setup).
Use this when the deployment can't reach `api.resend.com`, or you already have an internal mail relay (Exchange, Postfix, on-prem SendGrid, etc.). `SMTP_HOST` takes priority over Resend when both are set.
```bash
SMTP_HOST=smtp.internal.example.com
SMTP_PORT=587 # default 25; use 587 for STARTTLS submission
SMTP_USERNAME=multica # leave empty for unauthenticated relay
SMTP_PASSWORD=...
RESEND_FROM_EMAIL=noreply@yourdomain.com # reused as the From: header
```
Then restart: `docker compose -f docker-compose.selfhost.yml restart backend`.
For more auth configuration (OAuth, signup allowlist) and the full SMTP variable reference, see [Auth setup](/auth-setup) and [Environment variables → Email](/environment-variables#email-configuration).
## 4. First login + create a workspace
Open [http://localhost:3000](http://localhost:3000):
- Enter your email
- Grab the verification code from the Resend email (or, if you haven't configured Resend, from the server container stdout — look for the `[DEV] Verification code` line)
- Grab the verification code from your configured email backend (Resend or SMTP relay); if neither is configured, copy it from the server container stdout — look for the `[DEV] Verification code` line
- Do not use `888888` unless you explicitly set `MULTICA_DEV_VERIFICATION_CODE=888888` on a non-production private instance
- Log in and create your first workspace
## 5. Point the CLI at your own server
The CLI install is the same as in [Cloud quickstart → 2. Install the CLI](/cloud-quickstart#2-install-the-multica-cli) — Homebrew / script / PowerShell, pick one. Once installed, **use the self-host variant of the setup command**:
The CLI install is the same as in [Cloud quickstart → 2. Install the CLI](/cloud-quickstart#2-install-the-multica-cli) — Homebrew / script / PowerShell, pick one.
If you're running everything on one local machine:
If the CLI and the server run on the same host, the defaults already work:
```bash
multica setup self-host
```
That defaults to `http://localhost:8080` (backend) and `http://localhost:3000` (frontend).
That points the CLI at `http://localhost:8080` (backend) and `http://localhost:3000` (frontend), takes you through browser login, stores the PAT locally, and **starts the daemon automatically**.
`setup self-host` takes you through browser login, stores the PAT locally, and **starts the daemon automatically**.
### 5b. Cross-machine: front with a reverse proxy
Because the compose stack only listens on `127.0.0.1`, a daemon on a different machine cannot reach `http://<server-ip>:8080` directly — and you do not want it to, since the default `JWT_SECRET` would otherwise be reachable from the open internet. Put a reverse proxy on the server that terminates TLS and forwards to `127.0.0.1:8080` (backend) and `127.0.0.1:3000` (frontend), then point the CLI at the public HTTPS URL:
```bash
multica setup self-host \
--server-url https://<your-domain> \
--app-url https://<your-domain>
```
A minimal Caddyfile that fronts both the frontend and the backend (with WebSocket support, which the daemon and the web app both need) on a single hostname:
```nginx
multica.example.com {
# WebSocket route — must come before the catch-all
@ws path /ws /ws/*
handle @ws {
reverse_proxy 127.0.0.1:8080 {
flush_interval -1
}
}
# Backend API
handle /api/* {
reverse_proxy 127.0.0.1:8080
}
# Everything else → frontend
reverse_proxy 127.0.0.1:3000
}
```
After bringing the proxy up, set `FRONTEND_ORIGIN=https://multica.example.com` in the server's `.env` and restart the backend — otherwise the WebSocket origin check will reject the browser ([Troubleshooting → WebSocket can't connect](/troubleshooting#websocket-cant-connect)).
[Cloudflare Tunnel](https://developers.cloudflare.com/cloudflare-one/connections/connect-networks/) is another solid option — it gives you TLS and a public hostname without exposing any port on the host at all. An Nginx equivalent (separate `app.` / `api.` hostnames, `proxy_set_header Upgrade` for WebSockets) works just as well; the key requirements are TLS termination and forwarding the `Upgrade` header on `/ws`.
## 6. Create an agent + assign your first task
@@ -108,7 +158,7 @@ Same flow as Cloud — see [Cloud quickstart → Steps 5-6](/cloud-quickstart#5-
## Common issues
- **Backend won't start**: check container logs with `docker compose -f docker-compose.selfhost.yml logs backend`; usually it's a bad `DATABASE_URL` or `JWT_SECRET` in `.env`
- **Verification code not received**: Resend isn't configured → look for `[DEV] Verification code` in `docker compose logs backend`
- **Verification code not received**: no email backend is configured (neither Resend nor SMTP) → look for `[DEV] Verification code` in `docker compose logs backend`
- **WebSocket won't connect**: for public deployments you must set `FRONTEND_ORIGIN` to your real frontend domain; see [Troubleshooting → WebSocket won't connect](/troubleshooting#websocket-wont-connect)
description: "A squad is a group of agents (and optionally human members) led by one designated leader agent. Assign an issue to a squad and the leader decides who picks it up."
---
import { Callout } from "fumadocs-ui/components/callout";
A squad is a **named group of [agents](/agents) and human [members](/members-roles)**, with one designated **leader agent**. The squad is itself a first-class assignee: pick it from any **Assignee** picker and the leader takes the trigger, reads the issue, then `@`-mentions the squad member best suited to do the work. Squads let you assemble specialists once and dispatch them **by topic instead of by name** — the team grows, the routing stays the same.
## What a squad is, in mechanics
- **One leader, many members.** The leader must be an agent; members can be agents or human members. A squad with only the leader is allowed (the leader briefing notes "no other members"), and the same agent can sit in multiple squads.
- **Assignable everywhere a person is.** Squads appear in the Assignee picker, the @-mention picker, and the quick-create modal — anywhere you'd pick an agent or member, you can pick a squad.
- **Soft-deleted via archive.** Archive a squad and it disappears from pickers and lists; any issue currently assigned to it is **transferred to the leader agent** so the work doesn't go silent. Archived squads can't be assigned to new issues.
## When to use a squad versus a single agent
| Pick a squad when… | Pick a single agent when… |
|---|---|
| You have several specialists and don't know which one fits this issue in advance | The work is well-scoped to one specialty and you know who should do it |
| You want one stable assignee (the squad) while the actual responder changes per issue | You want the agent's name on the issue and clear individual accountability |
| You want a `@FrontendTeam` style routing target in comments | One-on-one `@agent-name` is enough |
The squad doesn't add capability — it adds **routing**. The members are still ordinary agents; the leader's only job is to pick the right one.
## Permissions
| Action | Who can do it |
|---|---|
| Create / update / archive a squad | Workspace **owner** or **admin** |
| Add or remove members, change roles | Workspace **owner** or **admin** |
| Assign an issue to a squad | Any workspace member (same as assigning to an agent) |
| `@`-mention a squad in a comment | Any workspace member |
| Record a squad-leader evaluation | The squad leader agent only (via CLI) |
The full role matrix lives in [Members and roles](/members-roles).
## Create a squad
In the sidebar, open **Squads → New squad** and fill in:
- **Name** — e.g. `Frontend Team`, `Bug Triage`. Doesn't need to be unique within the workspace.
- **Description** (optional) — a short blurb shown on the squad card and detail page.
- **Leader** — pick an existing agent. The leader is added to the squad automatically with role `leader`.
After creation, open the squad's detail page to:
- **Add members** — pick agents or human members, optionally give each a short role description (e.g. "owns the migrations", "reviewer of last resort"). The leader uses these roles when deciding who to delegate to.
- **Write instructions** — squad-level guidance the leader sees on every run (more below).
- **Set an avatar** — picked from the same picker used for agents.
When a non-Backlog issue is assigned to a squad, Multica immediately enqueues a `task` for the **leader agent** (not for every member). The flow then looks like this:
1. **Leader claims the task.** The agent runtime picks up the task on its next poll, same as any other agent assignment.
2. **Leader is briefed.** On claim, Multica appends three sections to the leader's system prompt — see [What the leader sees on every turn](#what-the-leader-sees-on-every-turn) below.
3. **Leader posts one delegation comment.** The comment `@`-mentions the chosen member(s) using the exact mention markdown from the roster — that mention triggers a new `task` for each mentioned agent.
4. **Leader records its evaluation** via `multica squad activity <issue-id> action --reason "..."`. This writes an entry to the issue's activity timeline so humans can see the leader actually evaluated the trigger.
5. **Leader stops.** The leader does not do the implementation itself. When the delegated member posts back, the leader is re-triggered to read the update and either delegate the next step, escalate, or stay silent.
If the issue is in **Backlog**, the leader is not triggered — Backlog is a parking lot, same rule as for direct agent assignment.
### What the leader sees on every turn
On each squad-leader run, three blocks are appended to the leader's instructions:
- **Squad Operating Protocol** — a hard-coded rule set: read the issue, delegate by `@`-mention, be terse (don't restate the issue body — the assignee can read it), record an evaluation every turn, and **stop after dispatching**. This protocol is system-managed and not editable.
- **Squad Roster** — the leader's self-row plus one row per non-archived member. Each row carries the exact mention markdown (`[@Name](mention://agent/<uuid>)` or `[@Name](mention://member/<uuid>)`) the leader should paste — typing a plain `@name` won't trigger anyone.
- **Squad Instructions** — your custom guidance for this squad (set on the squad detail page or via `multica squad update --instructions`). Use this for routing rules ("send DB work to Alice, frontend to Bob"), escalation policies, or anything else the leader needs to know that isn't already in the issue.
## When the leader is re-triggered
After the first dispatch, the leader is woken up automatically by **most subsequent comments** on the issue. The exact rules:
| Event | Leader triggered? |
|---|---|
| A non-member (human reporter, external agent) posts a comment | **Yes** |
| A squad member posts a progress update with no `@mention` | **Yes** — the leader re-evaluates whether the next step is needed |
| Anyone posts a comment that explicitly `@`-mentions another agent / member / squad / `@all` | **No** — the explicit `@` is the routing signal; the leader gets out of the way |
| The leader's own comment (self-trigger) | **No** — guarded to prevent a loop |
| A comment containing only an issue cross-reference (`[MUL-123](mention://issue/...)`) | **Yes** — issue references aren't routing |
Dedup applies on top of these rules: if the leader already has a `queued` or `dispatched` task on this issue, a new trigger won't enqueue a duplicate.
<Callout type="info">
**Why the leader doesn't trigger when a member posts an `@`-mention.** Once a squad member directly `@`s someone, that comment is a deliberate hand-off — having the leader wake up to "observe" the routing would just produce a no-op turn and clutter the timeline. Agent-authored comments are the exception: when an agent posts a result that `@`s another agent, the leader still wakes up so it can coordinate the thread.
</Callout>
## `@`-mention a squad in a comment
Squads appear in the `@` picker alongside members and agents. Mentioning a squad inserts `[@SquadName](mention://squad/<uuid>)` and triggers the **squad leader** as if you had assigned the issue to the squad — without changing the assignee or the status. Use this when you want the squad to pick someone for a question or sub-task while keeping the current owner.
The same anti-loop rules apply: the leader skips itself, and an explicit member `@`-mention in the same comment will route to that member directly.
## Reassign or archive a squad
**Reassigning an issue away from a squad** behaves like any other assignee change: all of the issue's active tasks (including the leader's) are cancelled, and the new assignee — agent, member, or another squad — is enqueued. There is no separate "remove squad without changing assignee" action; pick a different assignee.
**Archiving a squad** (`multica squad delete <id>`, or the Archive button on the detail page):
1. **Transfers issues currently assigned to the squad to the leader agent**, so the work continues against a concrete agent instead of going silent.
2. Marks the squad with `archived_at` / `archived_by` — the row is preserved so historical activity entries still resolve, but the squad disappears from lists, pickers, and the @-mention dropdown.
3. **Rejects future assignments** to this squad with `cannot assign to an archived squad`.
There is currently no unarchive command; create a new squad if you need the routing back.
## Squad operations from the CLI
| Command | Purpose |
|---|---|
| `multica squad list` | List squads in the workspace |
| `multica squad get <id>` | Show one squad's name, leader, description, instructions |
| `multica squad update <id> [--name X] [--description X] [--instructions X] [--leader Y] [--avatar-url Z]` | Update one or more fields |
| `multica squad delete <id>` | Archive (soft-delete) — transfers assigned issues to the leader |
| `multica squad member list <id>` | List a squad's members |
| `multica squad member add <id> --member-id <uuid> --type agent\|member [--role "..."]` | Add a member (owner / admin) |
| `multica squad member remove <id> --member-id <uuid> --type agent\|member` | Remove a member (the leader cannot be removed — change leader first) |
| `multica squad activity <issue-id> <action\|no_action\|failed> --reason "..."` | Recorded by the leader agent at the end of every turn |
`--leader` accepts an agent name or UUID; for everything else, IDs come from `multica agent list --output json`, `multica workspace members --output json`, and `multica squad list --output json`.
## Next
- [Assign issues to agents](/assigning-issues) — same flow, applies to squad assignees too
- [`@`-mention agents in comments](/mentioning-agents) — the `@` picker also surfaces squads
- [Agents](/agents) — what an agent is, the building block of every squad
- [Members and roles](/members-roles) — the full owner / admin / member permission matrix
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.