diff --git a/.beads/bd.sock.startlock b/.beads/bd.sock.startlock
deleted file mode 100644
index acd4cefc..00000000
--- a/.beads/bd.sock.startlock
+++ /dev/null
@@ -1 +0,0 @@
-64743
diff --git a/.beads/issues.jsonl b/.beads/issues.jsonl
index 84867083..44255229 100644
--- a/.beads/issues.jsonl
+++ b/.beads/issues.jsonl
@@ -278,7 +278,7 @@
 {"id":"bd-elqd","title":"Systematic bd sync stability investigation","description":"## Context\n\nbd sync has chronic instability issues that have persisted since inception:\n- issues.jsonl is always dirty after push\n- bd sync often creates messes requiring manual cleanup\n- Problems escalating despite accumulated bug fixes\n- Workarounds are getting increasingly draconian\n\n## Goal\n\nSystematically observe and diagnose bd sync failures rather than applying band-aid fixes.\n\n## Approach\n\n1. Start fresh session with latest binary (all fixes applied)\n2. Run bd sync and carefully observe what happens\n3. Document exact sequence of events when things go wrong\n4. File specific issues for each discrete problem identified\n5. Track the root causes, not just symptoms\n\n## Test Environment\n\n- Fresh clone or clean state\n- Latest bd binary with all bug fixes\n- Monitor both local and remote JSONL state\n- Check for timing issues, race conditions, merge conflicts\n\n## Success Criteria\n\n- Identify root causes of sync instability\n- Create actionable issues for each problem\n- Eventually achieve stable bd sync (no manual intervention needed)","status":"tombstone","priority":1,"issue_type":"task","created_at":"2025-12-16T22:57:25.35289-08:00","updated_at":"2025-12-17T16:11:17.070763-08:00","deleted_at":"2025-12-17T16:11:17.070763-08:00","deleted_by":"batch delete","delete_reason":"batch delete","original_type":"task"}
 {"id":"bd-epww","title":"Document pinned status in CLI_REFERENCE.md","description":"## Task\n\nThe `pinned` status is a valid issue status but is not documented in CLI_REFERENCE.md.\n\n## Current Status Values (from types.go:295-301)\n\n- open\n- in_progress\n- blocked\n- deferred\n- closed\n- tombstone\n- **pinned** ← undocumented\n\n## Definition\n\n`StatusPinned Status = \"pinned\" // Persistent bead that stays open indefinitely (bd-6v2)`\n\n## Where to Document\n\nAdd to the 'Issue Types' or create a new 'Issue Statuses' section in docs/CLI_REFERENCE.md.\n\n## Related\n\nThe pinned status is heavily used by gt for hook management (`--status=pinned`).","status":"closed","priority":2,"issue_type":"task","created_at":"2025-12-27T17:12:35.296002-08:00","created_by":"gastown/crew/joe","updated_at":"2025-12-27T19:13:05.640767-08:00","closed_at":"2025-12-27T19:13:05.640767-08:00","close_reason":"Added Issue Statuses section to CLI_REFERENCE.md documenting all 7 statuses including pinned"}
 {"id":"bd-etyv","title":"Smart --var detection for mol distill","description":"Implemented bidirectional syntax support for mol distill --var flag.\n\n**Problem:**\n- spawn uses: --var variable=value (assignment style)\n- distill used: --var value=variable (substitution style)\n- Agents would naturally guess spawn-style for both\n\n**Solution:**\nSmart detection that accepts BOTH syntaxes by checking which side appears in the epic text:\n- --var branch=feature-auth → finds 'feature-auth' in text → works\n- --var feature-auth=branch → finds 'feature-auth' in text → also works\n\n**Changes:**\n- Added parseDistillVar() with smart detection\n- Added collectSubgraphText() helper\n- Restructured runMolDistill to load subgraph before parsing vars\n- Updated help text to document both syntaxes\n- Added comprehensive tests in mol_test.go\n\n**Edge cases handled:**\n- Both sides found: prefers spawn-style (more common guess)\n- Neither found: helpful error message\n- Empty sides: validation error\n- Values containing '=' (e.g., KEY=VALUE): works via SplitN\n\nEmbodies the Beads philosophy: watch what agents do, make their guess correct.","status":"closed","priority":2,"issue_type":"task","created_at":"2025-12-21T11:08:50.83923-08:00","updated_at":"2025-12-21T11:08:56.432536-08:00","closed_at":"2025-12-21T11:08:56.432536-08:00","close_reason":"Implemented"}
-{"id":"bd-exy3","title":"Ephemeral patrol molecules leaking into beads","description":"## Problem\n\nEphemeral patrol orchestration molecules (e.g., mol-deacon-patrol, wisp step issues) keep appearing in `bd ready` output in gastown beads.\n\nThese are ephemeral/wisp issues that should:\n1. Never be synced to the beads-sync branch\n2. Live only in .beads-wisp/ (ephemeral storage)\n3. Be squashed to digests, not persisted as regular beads\n\n## Examples found\n\n- gt-wisp-6ue: mol-deacon-patrol\n- gt-pacdm: mol-deacon-patrol  \n- gt-wisp-mpm: Check own context limit\n- gt-wisp-lkc: Clean dead sessions\n\n## Investigation needed\n\n1. Where are these being created? (gt mol bond? manual bd create?)\n2. Why are they using the gt- prefix instead of staying ephemeral?\n3. Is the wisp storage (.beads-wisp/) being used correctly?\n4. Is bd sync accidentally picking up ephemeral issues?\n\n## Expected behavior\n\nPatrol molecules and their steps should be ephemeral and never appear in `bd ready` or `bd list`.","status":"open","priority":1,"issue_type":"bug","created_at":"2025-12-27T17:04:16.51075-08:00","created_by":"gastown/crew/joe","updated_at":"2025-12-27T17:04:16.51075-08:00"}
+{"id":"bd-exy3","title":"Ephemeral patrol molecules leaking into beads","description":"## Problem\n\nEphemeral patrol orchestration molecules (e.g., mol-deacon-patrol, wisp step issues) keep appearing in `bd ready` output in gastown beads.\n\nThese are ephemeral/wisp issues that should:\n1. Never be synced to the beads-sync branch\n2. Live only in .beads-wisp/ (ephemeral storage)\n3. Be squashed to digests, not persisted as regular beads\n\n## Examples found\n\n- gt-wisp-6ue: mol-deacon-patrol\n- gt-pacdm: mol-deacon-patrol  \n- gt-wisp-mpm: Check own context limit\n- gt-wisp-lkc: Clean dead sessions\n\n## Investigation needed\n\n1. Where are these being created? (gt mol bond? manual bd create?)\n2. Why are they using the gt- prefix instead of staying ephemeral?\n3. Is the wisp storage (.beads-wisp/) being used correctly?\n4. Is bd sync accidentally picking up ephemeral issues?\n\n## Expected behavior\n\nPatrol molecules and their steps should be ephemeral and never appear in `bd ready` or `bd list`.","status":"closed","priority":1,"issue_type":"bug","created_at":"2025-12-27T17:04:16.51075-08:00","created_by":"gastown/crew/joe","updated_at":"2025-12-27T19:18:16.809273-08:00","closed_at":"2025-12-27T19:18:16.809273-08:00","close_reason":"Moved to gastown: gt-glisw"}
 {"id":"bd-eyto","title":"Time-dependent tests may be flaky near TTL boundary","description":"Several tombstone merge tests use time.Now() to create test data: time.Now().Add(-24 * time.Hour), time.Now().Add(-60 * 24 * time.Hour), etc. While these work reliably in practice (24h vs 30d TTL has large margin), they could theoretically be flaky if: 1) Tests run slowly, 2) System clock changes during test, 3) TTL constants change. Recommendation: Consider using a fixed reference time or time injection for deterministic tests. Lower priority since current margin is large. Files: internal/merge/merge_test.go:1337-1338, 1352-1353, 1548-1549, 1590-1591","status":"open","priority":3,"issue_type":"task","created_at":"2025-12-05T16:37:02.348143-08:00","updated_at":"2025-12-05T16:37:02.348143-08:00"}
 {"id":"bd-f3ll","title":"Merge: bd-ot0w","description":"branch: polecat/dementus\ntarget: main\nsource_issue: bd-ot0w\nrig: beads","status":"closed","priority":2,"issue_type":"merge-request","created_at":"2025-12-19T23:20:33.495772-08:00","updated_at":"2025-12-20T23:17:27.000252-08:00","closed_at":"2025-12-20T23:17:27.000252-08:00","close_reason":"Branches nuked, MRs obsolete"}
 {"id":"bd-f526","title":"Deacon Patrol","description":"Mayor's daemon patrol loop for handling callbacks, health checks, and cleanup.","status":"open","priority":2,"issue_type":"molecule","created_at":"2025-12-27T18:14:20.809207-08:00","created_by":"deacon","updated_at":"2025-12-27T18:14:20.809207-08:00"}
diff --git a/.beads/sync-state.json b/.beads/sync-state.json
new file mode 100644
index 00000000..6fbc593b
--- /dev/null
+++ b/.beads/sync-state.json
@@ -0,0 +1,7 @@
+{
+  "last_failure": "2025-12-27T19:14:20.390485-08:00",
+  "failure_count": 1,
+  "backoff_until": "2025-12-27T19:14:50.390485-08:00",
+  "needs_manual_sync": false,
+  "failure_reason": "git pull failed in worktree: exit status 128\nFrom github.com:steveyegge/beads\n * branch              beads-sync -\u003e FETCH_HEAD\nfatal: Cannot rebase onto multiple branches.\n"
+}
\ No newline at end of file