majiayu000 Implement
Execute a plan by dispatching a fresh subagent per task. Each subagent implements, tests, self-reviews, and commits. No heavy review gates per task — the full review happens after all tasks complete (see dodi-dev:review).
Process
- Read the plan, extract all tasks with full text
- For each task sequentially: a. Dispatch implementer subagent with full task text + context (see implementer-prompt.md) b. Handle subagent status (see below) c. Mark task complete
- After all tasks: invoke
dodi-dev:review
Never dispatch multiple implementers in parallel — they'll conflict.
Model Selection
- Mechanical tasks (1-2 files, clear spec): use a fast model
- Integration tasks (multi-file, pattern matching): use standard model
- Architecture/judgment tasks: use most capable model
Handling Implementer Status
| Status | Action |
|---|---|
| DONE | Mark complete, next task |
| DONE_WITH_CONCERNS | Read concerns. If correctness issue, address first. If observation, note and proceed |
| NEEDS_CONTEXT | Provide missing context, re-dispatch |
| BLOCKED | Assess: provide more context, use more capable model, break task smaller, or escalate to human |
Never ignore an escalation or retry without changes.
Verification Along the Way
After each subagent completes, verify its claims before moving on:
- Check
git difforgit logto confirm changes were actually made - If the task included tests, confirm they actually ran and passed
- Don't trust "tests pass" without evidence