When agents use the multica CLI during task execution, their comments,
issue updates, and issue creations were attributed to the daemon's user
(via JWT) instead of the agent. Pass MULTICA_AGENT_ID env var from the
daemon, send X-Agent-ID header from the CLI client, and use it in
handlers to set the correct author/actor identity.
Previously, `multica login` only saved server_url to config when it was
empty. A stale value (e.g. from a prior session pointing at a different
host) was never overwritten, causing subsequent API calls to hit the
wrong server.
Main added execenv.Reuse() for workdir reuse across tasks on the same
issue. Our branch removed Type/BranchName/gitRoot from Environment
(repos are now checked out on demand). Resolution: keep Reuse() but
simplify it to work with the new Environment struct (no workspace type
tracking). Keep the "reused" log field from main, drop removed fields.
Agents now decide which repo to use based on issue context and check out
repos on demand via `multica repo checkout <url>`. Workspace repos are
cached locally as bare clones for fast worktree creation.
Key changes:
- Add repocache package for bare clone management (clone, fetch, worktree)
- Add `multica repo checkout` CLI command that talks to local daemon
- Add POST /repo/checkout endpoint on daemon health server
- Pass workspace repos metadata through register + task claim responses
- Remove pre-created worktrees from execenv (workdir starts empty)
- Update CLAUDE.md template to instruct agents to use `multica repo checkout`
- Pass MULTICA_DAEMON_PORT, WORKSPACE_ID, AGENT_NAME, TASK_ID env vars to agent
Previously each task created a fresh workdir via execenv.Prepare(), even
when resuming work on the same (agent, issue). This caused the agent's
session context to be out of sync with a blank code state.
Now the server returns prior_work_dir in the claim response, and the
daemon tries execenv.Reuse() first — which wraps the existing directory,
detects git worktree state, and refreshes context files. Falls back to
Prepare() if the prior workdir no longer exists. Workdirs are no longer
cleaned up after task completion so they remain available for reuse.
- Coalescing queue: use HasPendingTaskForIssue (queued/dispatched only)
instead of HasActiveTaskForIssue so comments during a running task
enqueue exactly one follow-up task that picks up all new comments.
- Stale task cleanup: runtime sweeper now fails orphaned tasks when
their runtime goes offline (daemon crash/network partition).
- Cancel-aware daemon: handleTask checks task status after execution
and discards results if the task was cancelled mid-run (e.g. reassign).
- Terminal issue guard: ClaimTaskForRuntime auto-cancels pending tasks
for done/cancelled issues instead of executing them.
- Race condition safety net: unique partial index ensures at most one
pending task per issue at the DB level.
member:added event now includes workspace_name. Frontend shows a toast
notification when the current user is invited. Also clears stale member
list on workspace switch.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Inbox events (new, read, archived, batch) are now sent via SendToUser
instead of broadcasting to the entire workspace room. Adds a new
Hub.SendToUser method. Also guards task broadcasts against deleted
issues to prevent global event leaks.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Inbox items were previously queried only by recipient, which leaked data
across workspaces. All list/count/batch operations now filter by
workspace_id from the X-Workspace-ID header.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
When a member comments on an issue assigned to an agent, automatically
enqueue a new task if the agent has on_comment trigger enabled (or no
triggers configured). Combined with session persistence, the agent
resumes its prior conversation context and sees the new feedback.
- Add HasActiveTaskForIssue query to prevent duplicate task enqueue
- Refactor shouldEnqueueAgentTask into reusable isAgentTriggerEnabled
- Add shouldEnqueueOnComment with active-task and status guards
- Call trigger logic from CreateComment handler
Add an `instructions` text field to the agent model, allowing users to
define each agent's role, expertise, and working style. Instructions are
injected into CLAUDE.md as an "Agent Identity" section so the agent
knows who it is on every task execution.
- Migration 021: add instructions column to agent table
- Backend: create/update/get agent handlers support instructions
- ClaimTask response includes instructions for daemon injection
- execenv: inject instructions into CLAUDE.md meta-skill
- Frontend: add Instructions tab to agent detail panel
Store the Claude Code session ID and working directory when a task
completes. On the next task for the same (agent, issue) pair, look up
the prior session and pass --resume <session_id> to Claude Code so
the agent retains conversation context across multiple tasks on the
same issue.
Changes:
- Migration 020: add session_id and work_dir columns to agent_task_queue
- CompleteAgentTask stores session_id and work_dir on completion
- GetLastTaskSession query retrieves prior session for (agent, issue)
- ClaimTaskByRuntime handler populates prior_session_id in response
- Daemon passes ResumeSessionID through to Claude backend Execute()
- Claude backend adds --resume flag when ResumeSessionID is set
Add per-workspace auto-incrementing issue numbers with a configurable
prefix, producing identifiers like "JIA-1" instead of truncated UUIDs.
Database:
- Add issue_prefix and issue_counter to workspace table
- Add number column to issue table with UNIQUE(workspace_id, number)
- Backfill existing issues with sequential numbers
Backend:
- Issue creation atomically increments counter in a transaction
- API responses include number and identifier fields
- Support issue lookup by identifier format (KEY-N)
- Workspace prefix auto-generated from name, customizable via API
Frontend:
- Display identifier in list rows and issue detail breadcrumb
- Add issue_prefix to Workspace type, number/identifier to Issue type
Add comprehensive data visualization to the runtime detail page:
- Daily token usage stacked area chart and daily cost bar chart
- Model distribution donut chart with cost breakdown
- GitHub-style activity heatmap (13 weeks of daily token usage)
- Hourly task distribution bar chart with new backend endpoint
- Responsive 2-column grid layout for charts on wide screens
Backend: new GET /api/runtimes/{runtimeId}/activity endpoint
returning hourly task counts from agent_task_queue.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The only path to marking a runtime offline was the daemon's deregister
call on graceful shutdown. If the daemon crashed, was killed, or lost
network, the status stayed "online" forever. Add a background goroutine
that sweeps every 30s and marks runtimes offline after 45s without a
heartbeat (3 missed intervals).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The createAgentInitIssue flow created a boilerplate "Initialize environment"
issue every time an agent was created, polluting the issue list with low-value
tasks. Agents naturally learn the project context when executing real work.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Refactor real-time sync from per-event precise mutations to WS-as-invalidation-signal + debounced refetch.
Backend:
- Add SubscribeAll to Event Bus — auto-broadcasts ALL events, eliminates manual 25-item allEvents list
- Add skill event constants to protocol, fix skill handler string literals
- Add title_changed activity tracking
Frontend:
- WSClient: add onAny() method for wildcard event subscription
- useRealtimeSync: rewrite to refreshMap + prefix routing + 100ms debounce
- Precise handlers only for side effects: workspace:deleted, member:removed, member:added (self-check)
- Reconnect now refetches all stores (fixes missing members/skills/workspace refresh)
- Stale-while-revalidate: fetch() only shows loading spinner on initial load, not on refetch
- Remove redundant useWSEvent in agents/page.tsx and skills-page.tsx
- WSClient.disconnect() now clears all handler registrations
Inbox bugfixes:
- Unify sidebar badge count with page count via dedupedItems + unreadCount in store
- Sort by time DESC (removed severity-first ordering)
- Ellipsis on truncated detail labels
UI:
- Status/Priority pickers: replace RadioGroup with MenuItem for auto-close on selection
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Add missing notifications for priority_changed and due_date_changed events
- Publish priority_changed and due_date_changed flags from UpdateIssue handler
- Add details JSONB column to inbox_item (migration 019) for structured change data
- Store from/to values in details for status, priority, assignee, and due_date changes
- Notification titles now use plain issue title; details carry structured context
- Add human-readable label maps (statusLabels, priorityLabels) in notification listeners
- Update inbox handler responses to include details field
- Frontend: InboxDetailLabel renders rich subtitles per notification type
- Status: "Set status to ● In Progress" with StatusIcon
- Priority: "Set priority to ◆ High" with PriorityIcon
- Assigned: "Assigned to Bob" with resolved actor name
- Due date: "Set due date to Apr 20"
- Comment: truncated comment body preview
- Frontend: HoverCard on inbox items shows issue title + description context
- Add due_date_changed to InboxItemType and typeLabels
- Add tests for priority_changed and due_date_changed notifications
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Extract shared timeAgo utility, remove duplicates from comment-card and issue-detail
- Remove unused replies prop from CommentCard
- Fix recursive delete to remove all descendant replies, not just direct children
- Improve formatActivity with human-readable status/priority labels and actor names
- Validate parent comment exists and belongs to same issue before creating reply
- Add priority_changed activity recording in activity listeners
- Fix activity SQL query to sort ASC (was DESC, then re-sorted in handler)
- Fix reply-input layout alignment and test submit button selector
- Minor: .gitignore additions, button dark mode aria-expanded fix
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Activity entries now show ActorAvatar instead of a small dot (consistent with comments)
- Remove All/Comments/Activity filter toggle (comments are just a type of activity)
- Remove one-level reply restriction in backend (allow nested threading)
- Remove unused Circle import
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Change parent_id FK from ON DELETE SET NULL to ON DELETE CASCADE.
Deleting a parent comment now deletes all its replies at the DB level.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Fix activity:created WS payload to match frontend expectations
(issue_id at top level, entry as TimelineEntry object)
- Promote child comments to top-level when parent is deleted
(both in handleDeleteComment and WS comment:deleted handler)
- Enforce one-level reply nesting: reject replies to replies with 400
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Replace the comment-only list with a Linear-style unified timeline that
interleaves field changes and comments chronologically.
Backend:
- activity_listeners.go: records field changes (status, assignee, description,
task completed/failed) to activity_log table on domain events
- Timeline API: GET /api/issues/{id}/timeline merges activity_log + comments
sorted by created_at
- Comment reply: parent_id column + handler support for threading
Frontend:
- Unified timeline replaces comment list: activity entries as compact muted
lines, comments as Card components with reply threading
- Filter toggle (All / Comments / Activity)
- Reply UI: inline editor under comments with Cancel/Reply buttons
- Real-time sync for activity:created + comment events
- 10 new Go tests, all passing
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Root cause: backend hardcoded UserType="member" in subscribe/unsubscribe handlers,
so unsubscribing an agent actually tried to delete a member record (no-op), and
the WS event broadcast wrong user_type causing other subscribers' UI to break.
- Backend: accept optional user_type in subscribe/unsubscribe request body
- Frontend: pass userType through API client to backend
- Fixes: clicking agent checkbox no longer affects member subscriptions
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Backend: subscribe/unsubscribe API now accepts optional user_id to manage other users
- Frontend: replace manual avatar stacking with shadcn AvatarGroup component
- Frontend: replace disabled-checkbox Popover with Command+Popover combo (Linear-style)
- Search/filter workspace members
- Click to toggle subscription (checkbox)
- Shows reason badge (Assignee, Creator, etc.) for auto-subscribed users
- Clean up duplicate status_change type (keep only status_changed)
- All tests pass (Go + TypeScript)
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Add comment explaining subscriber→notification listener registration order in main.go
- Add issue_status field to notifySubscribers and notifyDirect (fixes missing StatusIcon in inbox)
- Backfill existing commenters as subscribers in migration 016
- Add TODO comment for @mention duplicate notification prevention (deferred until @mention feature is enabled)
- Add context.Background() usage note for future bus-level timeout improvements
- Add toast error feedback on subscribe/unsubscribe failure
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Replace inbox_listeners.go with a subscriber-driven notification system:
- Add issue_subscriber table with auto-subscribe on create/assign/comment
- New subscriber_listeners.go: maintains subscriber data on domain events
- New notification_listeners.go: notifySubscribers (fanout to all subscribers
minus actor) and notifyDirect (targeted, punches through unsubscribe)
- Subscriber API: list/subscribe/unsubscribe endpoints
- Frontend: subscribers section in issue detail sidebar with real-time sync
- Frontend: inbox notification grouping by (issue_id, type, actor_id)
- Remove createInboxForIssueCreator from task.go (unified through event bus)
- 21 new Go tests, all passing
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Write skills to provider-native paths so agents discover them
automatically instead of relying on manual path references in
CLAUDE.md/AGENTS.md.
- Claude: write to {workDir}/.claude/skills/ (native discovery)
- Codex: write to per-task CODEX_HOME/skills/ with auth/config
seeded from ~/.codex/ (symlink auth.json, copy config files)
- Fallback: keep .agent_context/skills/ for unknown providers
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The parser read `payload.msg` but Codex JSONL files store token data at
`payload.info`. Also adds model tracking from `turn_context` events,
`last_token_usage` fallback, and `cache_read_input_tokens` field support.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* refactor: decouple task lifecycle from issue status, add daemon health server
- Remove automatic issue status changes from StartTask (in_progress),
CompleteTask (in_review), and FailTask (blocked) in task service.
Issue status is now fully managed by the agent via `multica issue status`.
- Update agent prompt and meta skill to instruct agents to manage issue
status themselves (in_progress → done/in_review/blocked).
- Add daemon health HTTP server on 127.0.0.1:19514 with /health endpoint
exposing pid, uptime, agents, and workspaces. Fail fast if port is taken
(another daemon already running).
- Update `multica status` to check both server and daemon health.
- Add Save button to repos section in workspace settings UI.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* refactor(daemon): simplify prompt, fix runtime config path, improve task error logging
- Slim down BuildPrompt to a minimal hint; detailed workflow now lives in CLAUDE.md/AGENTS.md
- Write CLAUDE.md to workDir root instead of .claude/CLAUDE.md
- Fix git-exclude pattern (.claude → CLAUDE.md)
- Decouple task queue reconciliation from issue status changes (agents manage status via CLI)
- Add diagnostic logging when CompleteTask/FailTask fail due to unexpected task state
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix(task): use task_completed/task_failed inbox notification types
FailTask was sending "agent_blocked" which conflates agent crash with
issue-level blocked status. Align notification types with the new
decoupled model: task_completed and task_failed. Update frontend types
and labels accordingly.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Add a `repos` JSONB column to the workspace table for storing
associated repository URLs and descriptions. This enables the daemon
to clone repos and set up git worktrees for agent task execution.
Structure: [{"url": "https://github.com/org/repo", "description": "..."}]
- Migration 014: adds repos column with default '[]'
- UpdateWorkspace query: supports updating repos
- Workspace API: returns repos in GET, accepts in PUT
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Replace the frozen context snapshot pattern with a CLI-driven approach:
agents now use `multica` CLI commands to fetch issue details, comments,
and workspace context on demand, always getting the latest data.
- Remove buildContextSnapshot and snapshot generation from enqueue
- Claim endpoint now returns fresh agent name + skills from DB
- Daemon resolves provider from local runtimeIndex, not snapshot
- Prompt instructs agent to use `multica issue get` / `comment list`
- Meta skill (CLAUDE.md/AGENTS.md) documents all available CLI commands
- Skills still injected as filesystem files (static agent config)
- Simplify daemon types: remove TaskContext/IssueContext/RuntimeContext
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Tests for truncateID, formatAssignee, resolveAssignee (6 cases),
and validIssueStatuses.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Make truncateID use rune counting instead of byte length for unicode safety
- Refactor workspaceGet and workspaceMembers to use newAPIClient helper
for consistent server-URL validation
- Add --output flag to issueStatusCmd for JSON output support
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Expose workspace details (including context field) and member listing
via the CLI so agents can dynamically query workspace context instead
of relying on static snapshots.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- resolveAssignee now reports actual API errors instead of silently
falling through to "not found" when both member/agent fetches fail
- Comment content truncation uses rune count for correct CJK handling
- Remove unnecessary _ = aType discard
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Add `multica issue` command group with subcommands for full issue
lifecycle management: list, get, create, update, assign, status,
and comment operations. Includes assignee resolution by name across
both workspace members and agents.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
These fields were unused in practice. Removed from frontend types,
issue detail UI, backend handlers, daemon prompt/context, protocol
messages, SQL queries, and tests. DB columns retained with defaults.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
