managing-beads

Managing Beads Issues

Session workflow

1. Run bd prime — loads context and shows open/claimed issues.
2. Pick work with bd ready, claim with bd update <id> --claim.
3. Work on code. Create follow-up issues as you discover them.
4. Commit issues.jsonl together with the code it tracks (the pre-commit hook exports it automatically).
5. Before context compaction: run bd sync to flush any pending writes.
6. Close finished issues with bd close <id>.

Core commands

bd prime                                                          # Session start: load context
bd ready                                                          # List unclaimed available issues
bd show <id>                                                      # View issue details
bd update <id> --claim                                            # Claim an issue
bd update <id> --status=in-progress                               # Update status
bd create --title="..." --description="..." --type=task --priority=2  # Create an issue
bd close <id>                                                     # Mark issue done
bd sync                                                           # Flush pending writes (before compaction)
bd bootstrap                                                      # Rebuild local DB after git pull
bd github sync                                                    # Bidirectional sync with GitHub Issues
bd github sync --pull-only                                        # Pull from GitHub only
bd github sync --push-only                                        # Push to GitHub only

Creating issues

bd create \
  --title="Add scroll support for web" \
  --description="CDP-based scroll not yet implemented; fdb scroll fails on web targets." \
  --type=task \
  --priority=2

Types: task, bug, feature, chore
Priorities: 1 (critical) → 4 (low)

Rules

• On main: git add .beads/issues.jsonl && git commit directly — the pre-commit hook exports to the right place.
• Always commit issues.jsonl together with the code changes it tracks — never in isolation.
• Run bd bootstrap after git pull on any machine that already has a local DB.
• Run bd prime at session start only — not during compaction (bd sync handles compaction).
• Never run bd doctor --fix — it can corrupt the local DB.
• bd is the source of truth; GitHub Issues are a mirror — sync with bd github sync.
• Capture any work discovered during implementation as new issues immediately.