The witness template previously instructed using raw git commands:
- git worktree remove polecats/<name>
- git branch -d polecat/<name>
This failed because git worktree operations must run from the repo
that created the worktree (mayor/rig or .repo.git), not the rig root.
Now uses `gt polecat nuke` which handles repo base detection via
the repoBase() function and properly cleans up worktrees.
Fixes gt-gp6i
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Added multiple layers of protection against circular redirects:
1. ResolveBeadsDir now detects when a redirect points back to itself
and auto-removes the errant redirect file with a warning
2. ensureBeadsRedirect now includes safety checks:
- Prevents creating redirects inside mayor/rig/ (the canonical location)
- Validates that the redirect target is not the same as the beads dir
3. Added test case for circular redirect detection
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
When GT_ROLE env var is not set, detectSender() now falls back to
cwd-based detection instead of immediately returning "overseer".
This allows running mail commands from witness/refinery directories
without GT_ROLE set (e.g., debugging sessions) and correctly detecting
the identity from the current working directory path.
(gt-od1g)
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
The dispatch command was documented in help text but never implemented.
Now it:
- Finds unassigned ready tasks in an epic
- Locates idle polecats (no hooked work)
- Slings the first available task to the first idle worker
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
When called with 1 arg from an agent working directory, gt mol attach
now auto-detects the pinned bead ID from the current agent's hook.
Uses the same role detection infrastructure as gt mail and bd. (gt-t7ekm)
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Root cause: gt done and mq submit created merge-request beads but did
not write to the .beads/mq/ directory that the refinery Engineer
polls for work.
Changes:
- done.go: Submit to mrqueue after creating MR bead
- mq_submit.go: Submit to mrqueue after creating MR bead
- mq_migrate.go: New command to migrate existing stale MR beads
- mrqueue.go: Follow beads redirects for shared queue location
The migration command allows recovery of existing stale MRs that were
never processed because they only existed as beads.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Previously gt prime only detected the Mayor role from town root (~/gt) or
~/gt/mayor/. Now it also detects Mayor from <rig>/mayor/ paths like
~/gt/gastown/mayor/rig, enabling mayor sessions to work correctly
regardless of which rig directory they're attached to.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
The mq list command was using r.Path (rig root) instead of r.BeadsPath()
which returns the mayor/rig clone path where beads are git-synced. This
caused the command to return empty results because it was looking at the
wrong .beads/ location.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Implement activity event emission for witness patrol operations:
- patrol_started: When witness begins patrol cycle
- polecat_checked: When witness checks a polecat
- polecat_nudged: When witness nudges a stuck polecat
- escalation_sent: When witness escalates to Mayor/Deacon
- patrol_complete: When patrol cycle finishes
Also adds refinery merge queue events for future use:
- merge_started, merge_complete, merge_failed, queue_processed
New command: `gt activity emit <event-type>` allows agents to emit
events from CLI. Events write to ~/gt/.events.jsonl for gt feed.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Add merge queue activity events to the refinery:
- merge_started: When refinery begins processing an MR
- merged: When MR successfully merged to main
- merge_failed: When merge fails (conflict, tests, push, etc.)
- merge_skipped: When MR skipped (superseded)
Events include MR ID, worker, branch, and reason (for failures).
Implemented in both Manager.ProcessMR and Engineer.ProcessMRFromQueue.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
The swarm status was stale because bd commands were running from the rig
root path instead of the mayor/rig clone which has proper beads sync config.
Changes:
- Add Rig.BeadsPath() method to return mayor/rig path when available
- Update all bd commands in swarm.go to use BeadsPath()
- Update swarm manager to use separate beadsDir and gitDir paths
- beadsDir for bd commands (git-synced location)
- gitDir for git operations and mail (rig root)
This ensures swarm status reflects the latest beads data from the
git-synced beads-sync branch instead of stale local daemon data.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Previously, UpdateAgentState embedded hook_bead in the description
text field, while bd slot commands read/write from the SQLite
hook_bead column. This caused gt sling to report hooks as occupied
when bd slot show showed them empty.
Fix: Change UpdateAgentState to use proper bd commands:
- `bd agent state` for agent_state (updates column directly)
- `bd slot set/clear` for hook_bead (updates column directly)
This ensures consistency between gastown and beads commands.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Implements `gt callbacks process` for Mayor/Deacon patrol to handle
callbacks from agents:
- POLECAT_DONE: Log polecat completion
- MERGE_COMPLETED: Close source issue, log merge
- MERGE_REJECTED: Log rejection reason
- HELP/ESCALATION: Forward to overseer
- SLING_REQUEST: Log sling request for spawn
- WITNESS_REPORT: Log rig health status
- REFINERY_REPORT: Log queue status
Also adds EventCallback to townlog for callback processing events.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Agent session molecules (gt-gastown-crew-joe, gt-gastown-witness, etc.)
update frequently for status tracking, creating noisy entries in the
activity feed. This change:
- Adds IsAgentSessionBead() to identify agent session beads
- Filters out "update" events for agent sessions from the event feed
- Still updates the agent tree so status is visible there
- Still shows create/complete/fail/delete events for agents
The filtering happens in addEvent() in the TUI feed model. Agent session
updates are identified by parsing the bead ID pattern and checking for
known agent roles (mayor, deacon, witness, refinery, crew, polecat).
Resolves: gt-sb6m4
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Phase 1 of activity feed improvements: gt commands now log events to
~/gt/.events.jsonl. This is the raw audit log that the feed daemon
(phase 2) will curate into the user-facing feed.
Instrumented commands:
- gt sling: logs sling events with bead and target
- gt hook: logs hook events with bead
- gt handoff: logs handoff events with subject
- gt done: logs done events with bead and branch
- gt mail send: logs mail events with to and subject
Event format follows the specification:
```json
{"ts":"2025-12-30T07:36:28Z","source":"gt","type":"mail",
"actor":"gastown/crew/joe","payload":{...},"visibility":"feed"}
```
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Adds a new workflow formula for polecat work that includes strategic
context reading before design. The context_docs variable allows
customization of which organizational documents to read.
Default: ~/gt/docs/hop/CONTEXT.md,~/gt/docs/PRIMING.md (Gas Town docs)
Override: gt sling formula --var context_docs="docs/ARCH.md,docs/STYLE.md"
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
When using `gt sling <formula> --on <bead>`, the command now properly
instantiates the formula as a molecule and bonds it to the target bead:
1. Cook the formula (ensures proto exists)
2. Create wisp with feature variable from bead title
3. Bond wisp to original bead (creates compound)
4. Hook the compound root (not bare bead) to target agent
This enables the "shiny" workflow (design→implement→review→test→submit)
to be applied to existing bugs/features, ensuring polecats complete all
phases including the code review step.
Example:
gt sling shiny --on gt-abc gastown
# Creates shiny molecule, bonds to gt-abc, slings compound to polecat
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
This fix addresses the issue where polecat sessions terminate unexpectedly
during work without recovery:
Changes:
- Add `checkPolecatSessionHealth()` to daemon heartbeat loop
- Proactively validates tmux sessions are alive for polecats
- Detects crashed polecats that have work-on-hook
- Auto-restarts crashed polecats with proper environment setup
- Notifies Witness if restart fails as fallback
- Add polecat support to lifecycle identity mapping
- `identityToSession()` now handles polecat identities
- `restartSession()` can restart crashed polecat sessions
- `identityToStateFile()` handles polecat state files
- `identityToAgentBeadID()` handles polecat agent beads
- `identityToBDActor()` handles polecat BD_ACTOR conversion
- Add `gt session check` command for manual health checking
- Validates tmux sessions exist for all polecats
- Shows summary of healthy vs not-running sessions
- Useful for debugging session issues
This provides faster recovery (within heartbeat interval) compared to
waiting for GUPP violation timeout (30 min) or Witness detection.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
