do-worklisted

# Do Work Output "Read Do Work skill." to chat to acknowledge you read this file. Pipeline position: `/grill-me` → `/write-a-prd` → `/architect` → `/prd-to-issues` → **`/do-work`** → `shft` ## Workflow ### 0. HUD event (if daemon running) Emit a session-start event so the HUD tracks this work: ```bash source ~/dotfiles/bin/write-hud-state.sh write_hud_event "info" "do-work: started — $TASK_SUMMARY" ``` Replace `$TASK_SUMMARY` with a short description. Emit again after each commit (`do-work: committed — <message>`) and at session end (`do-work: completed`). ### 1. Understand Read any referenced plan, PRD, or GitHub issue. If none provided, clarify the task with the user before proceeding. ### 2. Plan (optional) If the task has not already been planned, create a plan for it. Break large tasks into vertical slices (tracer bullets) — each slice should touch all layers end-to-end rather than building layer by layer. Skip this step if a plan or PRD already exists. ### 3. Implement Read a sample existing file of the same type before creating new ones — follow the conventions already established in the codebase. **For backend code**: use red/green/refactor, one test at a time in a tracer-bullet style. 1. Write a single failing test for the smallest vertical slice of behaviour 2. Run the test — confirm it fails (red) 3. Write the minimum code to make it pass (green) 4. Repeat from step 1 for the next slice 5. Refactor if needed while keeping tests green Do not write all