Track multi-agent task units as beads in a database with JSONL export.
npx skills add https://github.com/azigler/dotfiles --skill beads-azigler
Please help me install this Agent Skill. Skill: beads Source: https://github.com/azigler/dotfiles/tree/main/agents/skills/beads Command: npx skills add https://github.com/azigler/dotfiles --skill beads-azigler
This Skill solves the problem of keeping complex, multi-agent work organized so the right next action is always clear, with a complete handoff packet tied to code commits.
Use the beads skill to create and close one tracked task end-to-end by running: br create -p 2 "task: auth token refresh" then br update <id> --description "..." --acceptance-criteria "..." and finally br close <id> before you commit.
High-intent search queries and answers about installing and using this skill.
Multi-agent task tracking with commit-level traceability is managed by representing each unit of work as a bead persisted in an agent-friendly SQLite database. Beads are closed before merges and recorded in commit history via a Bead trailer for full traceability.
A bead handoff packet is a structured record containing description, design, acceptance criteria, notes, dependencies, and external references. It allows another agent to pick up a task cold without guessing by providing all required context in a standardized format.
Create a tracked task by running a CLI create command with a priority and title, update its structured fields using an update command, then close it before committing. Closing syncs the SQLite database state to JSONL for version control.
SQLite and JSONL exports support dependency-aware task triage by allowing commands to discover what work is ready, blocked, stale, or orphaned. The JSONL export enables version control integration while the database handles readiness checks across spec, check, test, impl, and eval stages.
This task tracking approach is designed specifically for multi-agent orchestration across spec, check, test, impl, and eval stages. It applies dependency-aware readiness checks and structured handoffs so agents can coordinate work transitions without losing context.
File-based task tracking without a database can lose dependency-aware readiness state and make triage across multiple agents difficult. This Skill addresses those limitations by persisting bead state in SQLite and syncing to JSONL for version control compatibility.
FindMalekConfigure Prisma for multiple database providers and set up the client.
RhettArkOrchestrate multi-agent workflows with the OpenAI Agents SDK in Python.
aziglerAudit bead populations and update JSONL state with lifecycle decisions.
DicklesworthstoneCoordinate interchangeable agents through bead-driven task plans in multi-agent workflows.
