Subagent Driven Development

Core Concept

FRESH AGENT = FRESH CONTEXT
Use subagents for focused work without baggage

When to Use Subagents

How to Launch Subagent

The Handoff:

1. DEFINE clear scope - what exactly to do
2. PROVIDE needed context - files, constraints, requirements
3. SPECIFY deliverable - what to return
4. SET success criteria - how to know it's done

Template:

Task: [Specific action to take]

Context:
- Working in [repo/directory]
- Relevant files: [list]
- Constraints: [any limits]

Deliverable:
- [Exact output expected]

Success when:
- [Criteria 1]
- [Criteria 2]

Anti-Patterns

❌ Vague handoffs

Bad: "Look into this bug"
Good: "Find root cause of TypeError in user.ts:45"

❌ Context dumping

Bad: Passing entire conversation history
Good: Passing only relevant files and specific question

❌ No success criteria

Bad: "Make it better"
Good: "Refactor to reduce duplication, all tests must pass"

❌ Too broad scope

Bad: "Implement the whole feature"
Good: "Implement the validation logic for email field"

Good Subagent Tasks

✅ Research:

• "Find how X is implemented in this codebase"
• "Search for similar patterns in the repo"
• "Understand the data flow from A to B"

✅ Focused fixes:

• "Fix the specific test failure in X"
• "Resolve the lint error in file Y"
• "Update function Z to handle edge case"

✅ Generation:

• "Generate tests for this function"
• "Create documentation for this module"
• "Write migration for schema change"

✅ Analysis:

• "Analyze dependencies of this module"
• "Identify all usages of this API"
• "Review this PR for issues"

Receiving Subagent Results

When subagent returns:

1. READ the full response
2. VERIFY against success criteria
3. INTEGRATE results into main context
4. CONTINUE from where you left off

Integration with Checkers

Before launching subagent:

• scope-boundary-checker → Is scope clear and bounded?
• assumption-checker → Are handoff assumptions valid?

After receiving results:

• fact-checker → Verify subagent claims
• pre-action-verifier → Before using results

Context Management

Main agent responsibilities:

• High-level plan
• User communication
• Final integration
• Overall progress

Subagent responsibilities:

• Focused execution
• Detailed work
• Return clean results
• No side conversations