bd sync: 2025-12-29 23:34:59

.beads/issues.jsonl

@@ -96,7 +96,7 @@
 {"id":"gt-1py3y","title":"Unify gt sling and gt spawn with --args support","description":"## Summary\n\nComplete the unification: `gt sling` becomes THE way to assign work. Remove top-level `gt spawn`.\n\n## Current State (Partially Done)\n\n- `gt sling` auto-spawns polecats when target is a rig\n- `gt sling --args` stores args in bead (no-tmux mode, gt-vc3l4)\n- `gt spawn` still exists with flags not yet in sling\n\n## Remaining Work\n\n### 1. Move spawn-only flags to sling\n\n| Flag | Current | Action |\n|------|---------|--------|\n| `--naked` | spawn only | Move to sling |\n| `--create` | spawn only | Move to sling |\n| `--molecule` | spawn only | Move to sling |\n| `--force` | spawn only | Move to sling |\n| `--account` | spawn only | Move to sling |\n| `--no-start` | spawn only | Evaluate if needed |\n\n### 2. Remove top-level gt spawn\n\n- Keep `gt spawn pending` as subcommand if still needed\n- Remove `gt spawn` from root commands\n- Update all docs and CLAUDE.md references\n\n### 3. Unified UX Design\n\nTarget resolution (smart defaults):\n```bash\ngt sling gt-abc                    # Self (current agent)\ngt sling gt-abc crew               # Crew worker in current rig  \ngt sling gt-abc gastown            # Auto-spawn polecat in rig\ngt sling gt-abc gastown/Toast      # Specific polecat\ngt sling gt-abc mayor              # Mayor\n```\n\nKey flags:\n```bash\ngt sling gt-abc --args \"patch release\"   # Natural language context\ngt sling gt-abc --naked                  # No-tmux mode (manual start)\ngt sling gt-abc --create                 # Create polecat if missing\ngt sling gt-abc --molecule mol-review    # Use specific workflow\n```\n\n### 4. Update Witness/Refinery/Deacon\n\nThese agents currently use `gt spawn`. Update to use `gt sling`:\n- Witness: spawning polecats for work\n- Refinery: spawning merge workers  \n- Deacon: any spawn logic\n\n## Acceptance Criteria\n\n- [ ] All spawn flags available in sling\n- [ ] `gt spawn` removed (or hidden/deprecated)\n- [ ] Witness uses `gt sling \u003crig\u003e` for polecat work\n- [ ] Docs updated\n- [ ] Both human and agent UX is clean","status":"tombstone","priority":1,"issue_type":"feature","created_at":"2025-12-26T13:35:55.902844-08:00","updated_at":"2025-12-27T21:29:45.90011-08:00","deleted_at":"2025-12-27T21:29:45.90011-08:00","deleted_by":"daemon","delete_reason":"delete","original_type":"feature"}
 {"id":"gt-1qti","title":"Mayor restart loop: auto-prime and mail check on attach","description":"When mayor session cycles (after /exit, gt mayor attach reconnects), the new session lacks context.\n\n## Current Behavior\n- Mayor exits, shell script loop restarts claude\n- New session starts cold - no gt prime, no mail check\n- Mayor is disoriented, doesn't know prior context\n- Strong 'antimemetic properties' - we discuss fixing it, then forget\n\n## Expected Behavior\nAfter restart, mayor should automatically:\n1. Run gt prime (load role context)\n2. Check gt mail inbox (find handoff messages)\n3. Look for 🤝 HANDOFF messages from predecessor\n\n## Possible Fixes\n1. Startup hook in Claude Code settings that runs gt prime\n2. CLAUDE.md instructions that say 'FIRST THING: run gt prime'\n3. Shell script wrapper that injects context before attach\n4. Modify gt mayor attach to inject prime command\n\n## Related\n- gt-vci: Mayor handoff mail template\n- gt-sye: Mayor startup protocol prompting","status":"tombstone","priority":1,"issue_type":"bug","created_at":"2025-12-19T14:44:43.188588-08:00","updated_at":"2025-12-27T21:29:54.025195-08:00","deleted_at":"2025-12-27T21:29:54.025195-08:00","deleted_by":"daemon","delete_reason":"delete","original_type":"bug"}
 {"id":"gt-1r6o6","title":"Digest: mol-deacon-patrol","description":"Patrol 17: All healthy","status":"closed","priority":2,"issue_type":"task","created_at":"2025-12-26T17:48:31.845975-08:00","updated_at":"2025-12-26T17:48:31.845975-08:00","closed_at":"2025-12-26T17:48:31.845924-08:00"}
-{"id":"gt-1rxz5","title":"Swarm status doesn't update after task completions","description":"During swarm bd-784c, the swarm status remained at '9% (1/11 tasks merged)' even after 5 tasks were completed and closed.\n\n**Expected:**\nSwarm status should reflect actual completion count, either by:\n1. Polling issue status from beads\n2. Updating when polecats close issues\n3. Updating when commits are pushed to integration branch\n\n**Impact:**\nCoordinator had to manually check bd list to know actual progress.","status":"hooked","priority":2,"issue_type":"bug","assignee":"gastown/polecats/rictus","created_at":"2025-12-28T22:14:05.745286-08:00","created_by":"beads/crew/emma","updated_at":"2025-12-29T23:30:19.701253-08:00"}
+{"id":"gt-1rxz5","title":"Swarm status doesn't update after task completions","description":"During swarm bd-784c, the swarm status remained at '9% (1/11 tasks merged)' even after 5 tasks were completed and closed.\n\n**Expected:**\nSwarm status should reflect actual completion count, either by:\n1. Polling issue status from beads\n2. Updating when polecats close issues\n3. Updating when commits are pushed to integration branch\n\n**Impact:**\nCoordinator had to manually check bd list to know actual progress.","status":"in_progress","priority":2,"issue_type":"bug","assignee":"gastown/polecats/rictus","created_at":"2025-12-28T22:14:05.745286-08:00","created_by":"beads/crew/emma","updated_at":"2025-12-29T23:34:54.641531-08:00"}
 {"id":"gt-1s89v","title":"gt crew next: Return self when only window in cycle group","description":"## Problem\n`gt crew next --session 'gt-mayor'` returns exit code 1 when the Mayor is the only window in that cycle group.\n\n## Expected Behavior\nWhen an agent is the only member of its cycle group (or not in any configured cycle group), `gt crew next` should return the current session itself rather than failing.\n\n## Context\n- C-b n in Mayor's session triggers this command\n- Currently only rig crews have configured cycle groups\n- Mayor is not part of any crew cycle group, so it should just return itself\n\n## Fix\nIf no other windows exist in the cycle group, return the current session name instead of error.","status":"tombstone","priority":1,"issue_type":"bug","created_at":"2025-12-27T16:46:28.657641-08:00","created_by":"gastown/crew/max","updated_at":"2025-12-27T21:29:45.740079-08:00","deleted_at":"2025-12-27T21:29:45.740079-08:00","deleted_by":"daemon","delete_reason":"delete","original_type":"bug"}
 {"id":"gt-1su","title":"Spawn inject: Enter not submitted after multiline paste","description":"When gt spawn injects a multiline context (85+ lines), Claude Code shows \n'[Pasted text #1 +85 lines]' but the prompt is not submitted.\n\nThe tmux SendKeys function does include 'Enter' but it appears to not work\nfor long pasted text:\n\n```go\nfunc (t *Tmux) SendKeys(session, keys string) error {\n    _, err := t.run(\"send-keys\", \"-t\", session, keys, \"Enter\")\n    return err\n}\n```\n\nPossible fixes:\n1. Use tmux paste-buffer instead of send-keys for long text\n2. Send Enter separately after the text\n3. Use bracketed paste mode correctly\n\nReproduction:\n```bash\ngt spawn gastown/Nux --issue gt-u1j.13 --create\n# Session shows pasted text but waits at prompt\n```","status":"tombstone","priority":1,"issue_type":"bug","created_at":"2025-12-17T14:09:20.774203-08:00","updated_at":"2025-12-25T01:30:41.67682-08:00","deleted_at":"2025-12-25T01:30:41.67682-08:00","deleted_by":"batch delete","delete_reason":"batch delete","original_type":"bug"}
 {"id":"gt-1tpts","title":"E2E Swarm Test Epic","status":"closed","priority":3,"issue_type":"epic","created_at":"2025-12-29T17:58:35.474031-08:00","created_by":"gastown/polecats/rictus","updated_at":"2025-12-29T18:02:47.847412-08:00","closed_at":"2025-12-29T18:02:47.847412-08:00","close_reason":"E2E Test: All tasks complete, swarm landed"}