Imported from try-works/rlm-workflow-acp (skills/rlm-subagent/SKILL.md). Install upstream with npx skills add try-works/rlm-workflow-acp --skill rlm-subagent. Copyright stays with the author.

Subagent-Driven Execution with Fallback

This skill provides parallel subagent execution for RLM Phase 3 (Implementation), Phase 3.5 (Code Review), and Phase 4 (Testing) with automatic fallback to sequential mode when subagents are unavailable.

Trigger examples

• Parallelize Phase 3 across independent sub-phases
• Dispatch an implementer subagent for each SP
• Run a separate code-reviewer subagent before Phase 4
• Subagents aren't available; fall back to sequential mode

Quick Reference

TODO Discipline for Subagents

The Iron Law: NO SUBAGENT COMPLETION REPORT WITHOUT ALL TODOS CHECKED.

Subagent TODO Requirements

Each subagent (implementer, code-reviewer) MUST:

1. Include ## TODO section in their work output
2. Check off items as work progresses
3. Verify ALL items checked before reporting completion

Implementer Subagent TODO Template

## TODO

- [ ] Read and understand assigned SP specification
- [ ] Ask clarifying questions (if any)
- [ ] Write failing test (RED phase)
- [ ] Run test and verify failure
- [ ] Implement minimal code (GREEN phase)
- [ ] Run test and verify pass
- [ ] Refactor while keeping tests green
- [ ] Run integration tests
- [ ] Self-review against plan
- [ ] Document changes made
- [ ] Report completion to controller

Code Reviewer Subagent TODO Template

## TODO

- [ ] Read original plan (Phase 3)
- [ ] Read implementation summary (Phase 3)
- [ ] Review git diff (BASE_SHA..HEAD_SHA)
- [ ] Verify plan alignment
- [ ] Assess code quality
- [ ] Check TDD compliance
- [ ] Categorize issues (Critical/Important/Minor)
- [ ] Document positive findings
- [ ] Render verdict
- [ ] Report review completion

Controller TODO Management

The controller MUST:

1. Verify subagent outputs include completed TODOs
2. NOT accept completion reports with unchecked items
3. Request subagent to complete remaining items

Capability Detection

At start of Phase 3, detect subagent availability:

IF can invoke "agent" or "Task" tool -> Use Parallel Mode
ELSE -> Use Sequential Fallback Mode

Detection rule: If the platform provides a subagent/task primitive, use parallel mode; otherwise fallback to sequential.

Parallel Mode (Subagents Available)

Phase 3: Parallel Sub-Phase Implementation

Controller responsibilities:

1. Read locked Phase 2 TO-BE plan once
2. Extract all sub-phases (SP1, SP2, SP3...)
3. Determine dependencies (independent vs sequential)
4. Dispatch implementer subagent per independent SP
5. Two-stage review after each SP completion
6. Integration testing after all approved

Dispatch pattern:

// Parallel dispatch for independent SPs
await Promise.all([
  Task({ description: "Implement SP1", prompt: implementerPrompt(SP1) }),
  Task({ description: "Implement SP2", prompt: implementerPrompt(SP2) }),
  Task({ description: "Implement SP3", prompt: implementerPrompt(SP3) })
])

Two-stage review:

1. Spec Review: Verify SP requirements met (plan alignment)
2. Code Quality Review: Assess code quality, TDD compliance, standards

Phase 3.5: Code Review Subagent

Trigger: After Phase 3 implementation Action: Dispatch agents/code-reviewer.md subagent with:

• BASE_SHA and HEAD_SHA of changes
• Phase 2 TO-BE plan for alignment check
• Severity classification (Critical/Important/Minor)

Review loop: Issues found -> implementer fixes -> re-review

Phase 4: Parallel Testing

Required before dispatching tests:

• Audit 03-implementation-summary.md against 00-requirements.md and 02-to-be-plan.md
• Document mismatches and remediation/addenda in Phase 4 artifact

Dispatch pattern:

// Parallel test execution
await Promise.all([
  Task({ description: "Run unit tests", prompt: testPrompt("unit") }),
  Task({ description: "Run integration tests", prompt: testPrompt("integration") }),
  Task({ description: "Run E2E tests", prompt: testPrompt("e2e") })
])

Result aggregation:

• Collect results from all subagents
• Summarize pass/fail counts
• Identify any critical failures

Sequential Fallback Mode (No Subagents)

Trigger: Subagent capability check fails

Characteristics:

• Execute sub-phases sequentially in main agent context
• Extended self-review checklist per sub-phase
• Integration testing between each sub-phase
• Full context preservation

Fallback trigger flow:

**Subagent Check:** NOT AVAILABLE
**Reason:** [Tool not found / Platform limitation / User request]
**Action:** Using SEQUENTIAL fallback mode

Sequential Execution Pattern

For each SP in [SP1, SP2, SP3...]:
  1. Execute SP implementation
  2. Self-review against plan (extended checklist)
  3. Document in Phase 4 artifact
  4. Run integration tests
  5. Proceed to next SP

Sequential Review Pattern

Phase 3.5 equivalent:

• Main agent performs extended self-review
• Use comprehensive checklist from agents/code-reviewer.md
• Document findings in Phase 3.5 artifact

Subagent Prompts

Implementer Subagent

File: agents/implementer.md

Key requirements:

• Self-contained context (full SP text provided)
• TDD discipline enforcement
• Question-before-work protocol
• Self-review checklist before completion

Usage:

You are an Implementer Agent. Implement this sub-phase:

**SP Text:** [full sub-phase text]
**BASE_SHA:** [commit SHA]
**Context:** [relevant files]

Follow the process in agents/implementer.md

Code Reviewer Subagent

File: agents/code-reviewer.md

Key requirements:

• Plan alignment verification
• Code quality assessment
• Severity classification (Critical/Important/Minor)
• Clear verdict (Approved / Changes Required)

Usage:

You are a Code Reviewer Agent. Review this implementation:

**Plan:** [Phase 3 TO-BE]
**Git Range:** [BASE_SHA..HEAD_SHA]
**Implementation:** [Phase 4 summary]

Follow the process in agents/code-reviewer.md

Artifact Documentation

Phase 4 artifact must include:

## Pre-Test Implementation Audit
- Requirements alignment (`00-requirements.md`): [summary + evidence]
- Plan alignment (`02-to-be-plan.md`): [summary + evidence]
- Mismatches and remediation/addenda: [details]

## Execution Mode
- **Mode:** Parallel / Sequential
- **Subagents Used:** [names and counts]
- **Fallback Reason:** [if applicable]

## Sub-phase Results
- SP1: [status] - [subagent name or "main agent"]
- SP2: [status] - [subagent name or "main agent"]
...

## Review Results
- SP1 Review: [status] - [reviewer name or "self-review"]
- SP2 Review: [status] - [reviewer name or "self-review"]
...

Phase 3.5 artifact (if used):

## Review Scope
- Git range: [BASE_SHA..HEAD_SHA]
- Execution Mode: Parallel (subagent) / Sequential (self-review)
- Reviewer: [subagent name / self]

## Issues Found
- Critical: [count]
- Important: [count]
- Minor: [count]

## Verdict
[APPROVED / APPROVED WITH NOTES / CHANGES REQUIRED]

Phase 4 artifact must include:

## Pre-Test Implementation Audit
- Requirements alignment (`00-requirements.md`): [summary + evidence]
- Plan alignment (`02-to-be-plan.md`): [summary + evidence]
- Mismatches and remediation/addenda: [details]

## Execution Mode
- **Mode:** Parallel / Sequential
- **Test Suites:**
  - Unit: [subagent name] / Main agent
  - Integration: [subagent name] / Main agent
  - E2E: [subagent name] / Main agent

## Results Summary
- Total execution time: [X] minutes
- Estimated sequential time: [Y] minutes
- Speedup: [Z]x

Decision Tree

Starting Phase 4
  |
  v
Subagent available? --NO--> SEQUENTIAL MODE
  |                         - Execute SPs sequentially
  |                         - Extended self-review
  |                         - Document as sequential
 YES
  |
  v
3+ independent SPs? --NO--> SEQUENTIAL MODE
  |                         - Single sub-phase
  |                         - No parallelism benefit
 YES
  |
  v
PARALLEL MODE
- Dispatch implementer per SP
- Two-stage review
- Integration after all done

Quality Maintenance

Both modes enforce:

• TDD discipline
• Artifact locking
• Gate compliance
• Integration testing

Parallel Mode advantages:

• Fresh context per sub-phase
• Independent verification
• Faster execution (3-5x)

Sequential Mode advantages:

• No subagent dependency
• Simpler coordination
• Full context accumulation

Common Patterns

Pattern 1: Multi-Domain Feature

Scenario: Feature touches API, UI, and database

Phase 2 TO-BE plan:

## Sub-phases
- SP1: API changes (backend/)
- SP2: UI changes (frontend/)
- SP3: Database migration (migrations/)

Phase 4 execution:

• SP1, SP2, SP3 dispatched in parallel (independent domains)
• Each implementer works on isolated files
• Integration testing after all complete

Pattern 2: Sequential Dependencies

Scenario: SP2 depends on SP1 output

Phase 3 execution:

• Dispatch SP1 -> wait -> review -> approve
• Dispatch SP2 (with SP1 context) -> wait -> review -> approve
• Sequential, not parallel

Pattern 3: High-Risk Change

Scenario: Critical auth system change

Phase 3: Implement with TDD Phase 3.5: Mandatory code review Phase 4: Parallel testing (unit + integration + E2E)

Troubleshooting

Subagent fails to complete

• Check if prompt was self-contained
• Verify BASE_SHA was provided
• Review logs for errors
• Retry with clearer instructions

Integration tests fail after parallel SPs

• Check for file conflicts between SPs
• Verify no shared mutable state
• Review commit history for ordering issues

Sequential mode too slow

• Consider breaking into more granular SPs
• Check if platform supports subagents
• Verify tool detection is working

References

• Main workflow: SKILL.md
• Implementer subagent: agents/implementer.md
• Code reviewer subagent: agents/code-reviewer.md
• Artifact templates: references/artifact-template.md