| name | ndx-work |
| description | Pick up a task from the PRD and begin working on it |
| argument-hint | [task-id] |
Pick up a task from the PRD and begin working on it.
- Read
.rex/workflow.md for the project's execution workflow. Follow its instructions — they define the expected discipline for task execution (TDD, validation, commit conventions, etc.)
- If task-id provided, call
get_item (rex MCP). Otherwise call get_next_task (rex MCP)
- Read task details: title, description, acceptance criteria, parent chain
- For files mentioned in the task, use
get_file_info and get_imports (sourcevision MCP) to understand current state
- Use
get_zone (sourcevision MCP) for the relevant architectural zone
- Present a work plan: what needs to change, which files, what tests
- After user approves the plan, call
update_task_status (rex MCP) to mark as in_progress
- Implement the changes following the workflow discipline
- Run validation and tests as specified in the workflow
- Call
append_log (rex MCP) with what was done, decisions made, and issues encountered
- When done, use
update_task_status (rex MCP) to mark as completed
- Record the work in hench run history so it is auditable alongside
ndx work runs: run ndx hench record --task=<id> --status=completed --title="<task title>" --summary="<one-line summary>". Use --status=cancelled (or failed) instead if the task was not completed.
Assisted run, not a hench run. This skill drives the task directly through Claude Code, so — unlike ndx work — it does not spawn the hench agent and cannot capture token usage (Claude Code does not expose its own token consumption to the skill). The record written in step 12 is marked assisted with zero token usage: it makes the work visible and auditable in run history, but ndx usage will show no tokens for it. Tell the user this when you finish so the absence of token totals is expected, not a bug.