Taskwarrior Workflow — Project Manager¶
Use Taskwarrior to publish ready queues per agent, enforce dependency order, and shard the backlog when multiple agents work in parallel.
Steps¶
- Ingest sources: Refresh tasks from
docs/features/**,docs/issues/**,docs/workplan/**; update priorities/dependencies. - Export/Backfill:
- Export current Taskwarrior state:
just taskwarrior-export - Backfill from JSON (dry-run):
just taskwarrior-backfill - Apply backfill:
just taskwarrior-backfill-apply - Shard pools (when N agents run in parallel):
- Partition tasks into N pools by role/priority:
just taskwarrior-pools pools=<N>(writesoutputs/workplan/taskwarrior-pool-*.json). - Ensure no cross-pool dependency; reassign or reorder until pools are independent.
- Publish ready queues:
- For each pool/role: provide the ready filter (
task +role:<ROLE> status:pending -BLOCKED next) orjust taskwarrior-next role=<ROLE>. - Export per-pool JSON for clarity (see pool outputs).
- Monitor & adjust:
- Rebalance pools if one agent drains early.
- Update priorities/dates based on operator feedback and dependency changes.
Notes¶
- Keep pools dependency-free; if impossible, merge affected tasks into the same pool.
- Annotate tasks with source doc path and target date to maintain traceability.
- Provide agents a short “next up” list per role after each planning cycle.