
Task tracking for AI agents
Claude, Cursor, Codex, Gemini—same backlog. Handoffs that work. Reviews that catch bugs.
Built for AI-agent workflows
Structured Handoffs
Capture done/remaining/decisions/uncertain. Next session doesn't guess.
Session Isolation
Different sessions must review. Catches "works on my context" bugs.
Query-Based Boards
Organize work with TDQ queries. View as swimlanes in the monitor.
Dependency Graphs
Model dependencies. Critical-path finds optimal unblocking sequence.
Epics
Track large initiatives spanning multiple issues with tree visualization.
Live Monitor
Real-time TUI dashboard watching agent activity across sessions.
How it works
You build the backlog. Agents work through it autonomously—in parallel, with handoffs and enforced review.
td create "OAuth login" -p P1td start td-a1b2td handoff --done "..." --remaining "..."td review td-a1b2td approve td-a1b2td monitorWatch work happen in real-time
The live monitor shows current work, board swimlanes, and activity logs across all agent sessions.

Works with your agent
Any agent that can run shell commands works with td—coding, research, data pipelines, whatever.


