Subagent Driven Development

Subagent-Driven Development

Execute plan by dispatching fresh subagent per task, with two-stage review after each: spec compliance review first, then code quality review.

Core principle: Fresh subagent per task + two-stage review (spec then quality) = high quality, fast iteration

When to Use

• Have implementation plan? → Yes
• Tasks mostly independent? → Yes
• Stay in this session? → Yes → Use this skill

vs. Executing Plans:

• Same session (no context switch)
• Fresh subagent per task (no context pollution)
• Two-stage review after each task: spec compliance first, then code quality
• Faster iteration (no human-in-loop between tasks)

The Process

1. Read plan, extract all tasks with full text, note context, create todo list

2. For each task:

   • Dispatch implementer subagent with full task text + context
   • If subagent asks questions → answer, provide context, re-dispatch
   • Implementer implements, tests, commits, self-reviews

3. Dispatch spec reviewer subagent

   • Confirms code matches spec
   • If issues → implementer fixes → re-review

4. Dispatch code quality reviewer subagent

   • Reviews code quality
   • If issues → implementer fixes → re-review

5. Mark task complete

6. Repeat for remaining tasks

7. After all tasks: Use finishing-a-development-branch skill

Using runSubagent Tool

When dispatching subagents, use the runSubagent tool:

runSubagent(
  prompt: "Detailed task description with full context",
  description: "Brief 3-5 word summary"
)

For implementer:

runSubagent(
  prompt: "Implement Task 1: [full task text from plan]
  
  Context: [relevant project context]
  
  Requirements:
  - Follow TDD
  - Commit when done
  - Self-review before finishing
  
  Return: Summary of implementation and any issues found",
  description: "Implement [component name]"
)

For spec reviewer:

runSubagent(
  prompt: "Review implementation against spec.
  
  Task spec: [task requirements]
  
  Check:
  - All requirements met
  - Nothing extra added
  - Tests cover requirements
  
  Return: ✅ Spec compliant OR ❌ Issues: [list]",
  description: "Spec review task N"
)

For code quality reviewer:

runSubagent(
  prompt: "Review code quality of recent changes.
  
  Review for:
  - Clean code principles
  - Proper error handling
  - Test coverage
  - No obvious bugs
  
  Return: Strengths, Issues (by severity), Assessment",
  description: "Code review task N"  
)

Red Flags

Never:

• Skip reviews (spec compliance OR code quality)
• Proceed with unfixed issues
• Dispatch multiple implementation subagents in parallel (conflicts)
• Make subagent read plan file (provide full text instead)
• Skip scene-setting context
• Ignore subagent questions
• Accept "close enough" on spec compliance
• Skip review loops
• Start code quality review before spec compliance is ✅
• Move to next task while either review has open issues

If subagent asks questions:

• Answer clearly and completely
• Provide additional context if needed
• Don't rush them into implementation

If reviewer finds issues:

• Implementer (same subagent) fixes them
• Reviewer reviews again
• Repeat until approved
• Don't skip the re-review

Integration

Required workflow skills:

• writing-plans - Creates the plan this skill executes
• requesting-code-review - Code review template for reviewer subagents
• finishing-a-development-branch - Complete development after all tasks

Subagents should use:

• test-driven-development - Subagents follow TDD for each task