Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
Version History
- Made it less coding specific and more generic. - Updated documentation to clarify usage, examples, and conventions. - Improved language for brevity and readability. - Modernized sample task and epic names in command examples. - Noted that commit-message convention is optional. - Removed references to agent-incompatible workflows and commands (e.g. dropped mention of `git status` in the end-of-session checklist).
What This Skill Does
Distributed, git-backed issue tracker for AI agents. Stores tasks as JSONL in .beads/ with a dependency-aware graph that replaces markdown plans. Supports multi-agent coordination, hierarchical epics, and machine-readable output via --json flags.
Unlike markdown task lists, beads lets agents query unblocked work directly with structured JSON and enforces dependency ordering automatically.
When to use it
- Tracking multi-step agent tasks with blocking dependencies
- Coordinating parallel work across multiple AI agents on one project
- Querying the next unblocked task in a project work queue
- Organizing large projects into epics with nested subtasks
- Persisting task state across sessions via git commits
Example Workflow
Here's how your AI assistant might use this skill in practice.
User asks: Plan and track a multi-step coding project with an AI agent
- 1Run bd init --quiet to initialize the task database in .beads/
- 2Run bd create for each task with priorities and dependency links using --json
- 3Run bd ready --json to get the current unblocked work queue
- 4Run bd update <id> --status in_progress --assignee agent-1 to claim a task
- 5Run bd close <id> --reason "Done" then bd sync to commit and flush changes to git
Tasks stored in .beads/ as a dependency graph, committed to git, with remaining work surfaced for agent handoff
Requirements
Accounts, API keys, or tools you or your AI assistant may need to set up while using this skill.