Manages task ownership and progress tracking within LLM sessions to streamline multi-agent coordination.
BA (Task Tracker) is a lightweight, Git-friendly task management system specifically designed for LLM sessions and multi-agent development environments. It facilitates a structured workflow by allowing agents to discover available work, claim ownership of specific issues using unique session IDs, and track tasks through their entire lifecycle. By utilizing a line-based JSONL storage format and built-in dependency tracking (blocking/unblocking), BA ensures clear hand-offs and prevents redundant efforts when multiple AI agents or developers are working on the same codebase, making it an essential tool for complex, collaborative AI projects.
Key Features
01Dependency management with issue blocking and tree visualization
02JSON output support for programmatic integration with other developer tools
03Robust CLI for creating, claiming, transitioning, and documenting tasks
04Ownership-based state machine to prevent multi-agent work conflicts
05Git-friendly JSONL storage for easy version control and human readability
069 GitHub stars
Use Cases
01Visualizing task dependencies to determine the optimal order of operations for AI agents
02Managing a local backlog of refactors and technical debt directly in the repository
03Coordinating complex features across multiple concurrent Claude Code sessions
