beadboard/docs/prompts/2026-02-28-next-session-dolt-repair.md
zenchantlive c246ceaf21 feat(ux): consolidate Launch Swarm + telemetry UX with minimized strip
- Removed broken LaunchSwarmDialog (formula-based) from TopBar/LeftPanel
- All Rocket buttons (TopBar, LeftPanel, DAG nodes, social cards) now open
  AssignmentPanel (archetype-based) which actually works
- Every Rocket clears taskId first so assignMode && !taskId condition passes
- Conversation button priority: taskId always shows conversation, not assign panel
- Added TelemetryStrip: minimized right sidebar with status dots when non-telemetry
  panel (conversation/assignment) is active
- Live feed has minimize button → restores last taskId or assignMode
- DAG nodes: Signal icon → restores telemetry feed
- Social button on DAG nodes: single router.push to avoid race (setView + setTaskId)
- Fixed social card message button: opens right panel with drawer:closed (no popup)

Co-Authored-By: Oz <oz-agent@warp.dev>
2026-03-01 18:17:58 -08:00

75 lines
2.9 KiB
Markdown

# Next Session Prompt: Dolt/Beads Database Recovery and Source-of-Truth Reconciliation
You are continuing work in `/mnt/c/Users/Zenchant/codex/beadboard`.
## Problem snapshot (as of 2026-02-28)
Bead state appears empty in `bd ready`, but `.beads/issues.jsonl` contains historical data.
Observed signals:
- `bd status` reports only **3 closed issues** (0 open/in_progress/blocked).
- `.beads/issues.jsonl` has **381 lines**.
- `bd doctor --dry-run` reports:
- **Dolt-JSONL count mismatch** (`Dolt has 3, JSONL has 381`)
- stale/merge artifacts and lock issues
- config mismatch hints in `.beads/metadata.json`
- Prior repair attempt hit lock/permission conflicts (`database is locked by another dolt process`).
## Goal
Restore BeadBoard to a consistent, usable state where `bd` in this repo reflects the expected issue set, and `bd ready` shows correct ready work.
## Non-goals
- No frontend feature work.
- No broad refactors.
- No destructive git resets.
## Required outcomes
1. Clear diagnosis of why Dolt state diverged from JSONL state.
2. Safe repair path applied (or a precise blocker report if repair cannot complete).
3. Post-repair verification evidence showing issue counts and ready-state are sane.
4. Short write-up on recurrence prevention.
## Guardrails
- Treat `.beads/issues.jsonl` as historical source candidate, but verify before forcing import.
- Avoid direct manual edits to `.beads/issues.jsonl`.
- Use `bd` commands and documented repair flows first.
- Do not run destructive git commands.
## Suggested execution order
1. **Capture baseline evidence**
- `bd where`
- `bd status`
- `bd ready`
- `bd list --status open --status in_progress --status blocked --json`
- `wc -l .beads/issues.jsonl`
- `bd doctor --dry-run`
2. **Check lock/process state**
- Identify active `bd`/`dolt` processes.
- Identify stale lock files in `.beads`.
- Resolve lock conflicts in a non-destructive way.
3. **Repair using bd-supported flow**
- Attempt `bd doctor --fix --source=jsonl --yes`.
- If still divergent, run focused checks (`bd doctor --check=validate`, `bd doctor --check=artifacts`).
- If needed, document and execute the minimal additional `bd` recovery command sequence.
4. **Reconcile and verify**
- Re-run `bd status`, `bd ready`, `bd list --json`.
- Confirm counts are no longer `3 vs 381` divergent.
- Confirm expected open/ready beads appear.
5. **Record root cause + prevention**
- Add a concise note in docs (or bead notes) on:
- root cause
- repair steps that worked
- required hygiene (locks, hooks, sync flow, config keys)
## Deliverables
1. Repair summary with command evidence.
2. Final healthy `bd` status/ready output summary.
3. Minimal prevention checklist for future sessions.
## Completion criteria
- `bd` issue inventory in this repo is consistent with expected project history.
- `bd ready` is no longer falsely empty due to DB divergence.
- No destructive repository actions were used.