ワンクリックで
beads-workflow
// Use Beads (`bd`) as the durable task-management system for work that needs priorities, dependencies, status tracking, and cross-session continuity.
// Use Beads (`bd`) as the durable task-management system for work that needs priorities, dependencies, status tracking, and cross-session continuity.
Turn a one-line objective into a step-by-step construction plan where every step has a self-contained context brief. Designed for work that spans multiple sessions or agents.
Treat diagnosis as a first-class phase before fixing a bug, unstable behavior, or weird operational issue.
Use contracts, templates, and durable markdown artifacts consistently when a workflow needs handoffs, plans, context bundles, decisions, or verification evidence.
Use this when designing an internal tool or operational app where the safest path is to prove the workflow and UI before locking in backend structure.
Create a practical understanding of a repository or a large area before detailed work begins.
This is the meta-skill -- it teaches how to build skills that AI agents can
| name | beads-workflow |
| description | Use Beads (`bd`) as the durable task-management system for work that needs priorities, dependencies, status tracking, and cross-session continuity. |
| pack | skills-autonomous |
Use Beads (bd) as the durable task-management system for work that needs to survive session boundaries, compaction, and multi-agent handoffs. Beads stores tasks in a Dolt database (.beads/) with priorities, dependencies, status, and audit trails.
Use this skill when the user mentions:
BeadsbdDo not use this skill for a tiny one-shot checklist that can be handled entirely in the current session.
| Situation | Use This |
|---|---|
| Tracking what to do next | bd ready |
| Multi-step implementation plan | bd create for each task with dependencies |
| Quick in-session task list | OpenCode TodoWrite (ephemeral, fine for single sessions) |
| Architecture/design planning | /plan command + bd create for resulting tasks |
| Multi-session work | bd (persists) + /save-session (context snapshot) |
| Bug tracking / follow-up | bd create -t bug |
| Cross-agent coordination | bd update --claim (atomic assignment) |
Rule of thumb: If the work might span compaction or multiple sessions, use bd. If it's a quick checklist for the next 10 minutes, TodoWrite is fine.
At the start of every session on a project with Beads initialized:
# Check what's ready to work on (no open blockers)
bd ready --json
# If resuming, check what was in progress
bd list --status in_progress --json
Read the output. If there are in-progress tasks, resume them. If not, pick the highest-priority ready task.
Beads commonly uses Dolt-backed local state. In sandboxed agent sessions:
bd ready --json as a command that may need escalation.bd ready --json fails with Dolt server startup/connect errors, retry the same command outside the sandbox once.sandbox blocked from Beads unavailable:
sandbox blocked means retry outside the sandbox is needed or still fails for environment reasonsBeads unavailable means bd is not installed or .beads/ is not initializedCreate tasks from plans, user requests, or discovered work:
# Basic task
bd create "Implement user authentication" -p 1
# With type (task, bug, epic, message)
bd create "Fix login redirect loop" -t bug -p 0
# Epic with sub-tasks
bd create "Add export feature" -t epic -p 1
# Returns bd-xxxx
bd create "Design export API schema" -p 1 --parent bd-xxxx
bd create "Implement CSV export" -p 2 --parent bd-xxxx
bd create "Implement JSON export" -p 2 --parent bd-xxxx
bd create "Add export UI button" -p 2 --parent bd-xxxx
In multi-app or multi-business repos, do not leave feature-sized work as unlabeled flat tasks.
Preferred structure:
Examples:
app:listing-helperfeature:bulk-editortype:featurebusiness:spiffing-printsIf bv is part of the workflow, these labels should be treated as visibility keys, not optional decoration. Without app/feature labels, portfolio views will degrade into unmapped backlog summaries.
If app / feature / type metadata is missing at planning time, ask the user before creating the Beads structure.
| Priority | When |
|---|---|
| P0 | Blocking all progress, security issues, broken builds |
| P1 | Must be done this session / sprint |
| P2 | Should be done soon, not blocking |
| P3 | Nice to have, backlog |
# Claim a task (atomic — sets assignee + in_progress)
bd update bd-xxxx --claim
# Add dependencies (child is blocked by parent)
bd dep add bd-child bd-parent
# Add design notes
bd update bd-xxxx --design "Using repository pattern with async adapter"
# Close when done
bd close bd-xxxx --reason "Implemented and tested"
When /plan produces an implementation plan, convert it to Beads tasks:
/plan [description] — produces approach, task breakdown, risksbd create "[feature name]" -t epic -p 1bd ready --json to see what's unblockedWhen running multi-agent pipelines:
bd update --claimBefore ending a session:
# Close completed work
bd close bd-xxxx --reason "Completed"
# File issues for remaining work
bd create "Follow-up: add edge case tests" -p 2
# Check what's left
bd ready --json
# Save session context (our framework command)
/save-session [name]
Beads state survives compaction automatically (it's on disk). After compaction:
# Reload task context
bd ready --json
bd list --status in_progress --json
The session-memory hook should include a reminder to check bd ready after compaction.
If bd ready --json fails with Dolt startup/connect errors during recovery, retry outside the sandbox before giving up.
| Command | Purpose |
|---|---|
bd init | Initialize Beads in a project |
bd ready | List tasks with no open blockers |
bd create "Title" -p N | Create a task |
bd update ID --claim | Atomically claim a task |
bd close ID | Close a completed task |
bd show ID | View task details |
bd list | List all tasks |
bd dep add child parent | Add dependency |
bd prime | Load context summary for the agent |
Always use --json flag when parsing output programmatically.
bd ready. It handles dependency resolution — don't manually pick tasks.