Do Plan Skill
Execute implementation plans by deploying subagents for each phase. You coordinate; they execute.
When to Use
- After
/make-planhas created a structured plan - For any multi-phase task that needs coordinated execution
- When you want verified, incremental progress with commits
How It Works
You are an ORCHESTRATOR. Deploy subagents to execute all work.
Do not do the work yourself except to:
- Coordinate and route context
- Verify that each subagent completed its assigned checklist
- Decide whether to advance to the next phase
Execution Protocol
For Each Phase
Deploy an "Implementation" subagent to:
- Execute the implementation as specified in the plan
- COPY patterns from documentation - don't invent
- Cite documentation sources in code comments when using unfamiliar APIs
- If an API seems missing, STOP and verify - don't assume it exists
After Each Phase
Deploy subagents for verification (do not proceed until all pass):
| Subagent | Responsibility |
|---|---|
| Verification | Run the phase's verification checklist, prove it worked |
| Anti-pattern | Grep for known bad patterns from the plan |
| Code Quality | Review changes for obvious issues |
| Commit | Commit only after verification passes |
Between Phases
Deploy a "Branch/Sync" subagent to:
- Push to working branch after each verified phase
- Prepare the next phase handoff with plan context
Orchestrator Rules
- Each phase uses fresh subagents where noted (or when context is large/unclear)
- Assign one clear objective per subagent and require evidence:
- Commands run
- Outputs produced
- Files changed
- Do not advance until the assigned subagent reports completion AND you confirm it matches the plan
Failure Modes to Prevent
- ❌ Don't invent APIs that "should" exist - verify against docs
- ❌ Don't add undocumented parameters - copy exact signatures
- ❌ Don't skip verification - deploy a verification subagent
- ❌ Don't commit before verification passes
Example Usage
User: /do-plan plans/websocket-feature.md
You (orchestrator):
1. Read the plan file
2. For Phase 1:
a. Deploy Implementation subagent with phase context
b. Wait for completion report with evidence
c. Deploy Verification subagent
d. Deploy Anti-pattern subagent
e. Deploy Code Quality subagent
f. If all pass → Deploy Commit subagent
g. If any fail → Debug and redeploy
3. Repeat for each phase
4. Report overall completion status
Subagent Spawning
Use sessions_spawn to deploy subagents:
sessions_spawn(
task="[Phase N Implementation] Copy the auth middleware pattern from docs/auth.md:45-80 into src/middleware/websocket.ts. Report: files changed, commands run, verification status.",
label="phase-1-impl"
)
Each subagent should be given:
- Clear scope (one phase, one responsibility)
- Specific file/doc references from the plan
- Required output format (evidence of completion)
Progress Tracking
Keep a running status in the conversation:
## Execution Status
- [x] Phase 0: Documentation Discovery ✅
- [x] Phase 1: Core WebSocket Handler ✅ (commit: abc123)
- [ ] Phase 2: Authentication Integration (in progress)
- [ ] Phase 3: Error Handling
- [ ] Phase 4: Final Verification
