Agent Name

Purpose

What this agent does and when it should be invoked.

When You're Called

Task(
    subagent_type="your-agent-name",
    description="Brief description",
    prompt="Detailed instructions"
)

Process Flow

Phase 1: [Description]
     ↓
Phase 2: [Description] → VERIFY before proceeding
     ↓
Phase 3: [Description] → VERIFY before proceeding
     ↓
Report Summary

Phase 1: [Name]

[Detailed instructions for this phase]

Phase 2: [Name]

[Detailed instructions for this phase]

GATE 3 CHECKPOINT: Before proceeding to Phase 3, verify:

• Phase 2 completed successfully
• Expected outputs are present
• No errors that would invalidate Phase 3

Phase 3: [Name]

[Detailed instructions for this phase]

Critical Rules

Sequential Execution

• Process phases in order
• Wait for completion before calling next phase
• No parallel execution of dependent steps
• Each Task() call must complete before starting the next

Error Handling

• If a phase fails, note it but continue if possible
• Report all failures in final summary
• Never proceed if verification fails
• Track which sub-tasks succeeded and which failed

Expected Output

Phase Results:
✓ Phase 1: [Description] - Completed
✓ Phase 2: [Description] - Completed (X items processed)
✗ Phase 3: [Description] - Failed (reason)

Summary:
- Total phases: 3
- Completed: 2
- Failed: 1

Errors:
- Phase 3: [Specific error message]

Six Gates Compliance

Gate	Status	Notes
1. Read Before Write	✅	Reads all inputs before mutations
2. File Exists Check	✅	Validates targets in Phase 1
3. Chain Validation	✅	Checkpoints between phases
4. Mutation Confirm	✅	Reports preview before write
5. MCP Health	N/A	[Adjust based on MCP tool usage]
6. Post Validation	✅	Verifies writes in summary

Example Invocations

From Parent Agent

Task(
    subagent_type="your-agent-name",
    description="Process [something]",
    prompt="Process [details]"
)

From User (Manual)

Task(
    subagent_type="your-agent-name",
    description="Manual invocation",
    prompt="[User's request]"
)

Notes

• [Additional implementation notes]
• [Edge cases to be aware of]
• [Dependencies on other agents/tools]