Checkpoint: Multi-Agent System

Prove your multi-agent setup works.

The Challenge

Complete this three-part test:

Part 1: Spawn a Researcher

Spawn a researcher to answer:

"What are the top 3 note-taking apps for developers in 2024?"

Success criteria:

• Sub-agent spawns without error
• Results return within 3 minutes
• At least 3 apps are listed with descriptions
• Sources are cited

Part 2: Spawn an Executor

Spawn an executor to:

"Create a file called 'test-execution.md' in the workspace with today's date and a hello message"

Success criteria:

• Sub-agent spawns without error
• File is created
• File contains correct content
• Execution is confirmed

Part 3: Parallel Spawn

Spawn 2 researchers in parallel:

• One to research "best productivity apps"
• One to research "best note-taking apps"

Success criteria:

• Both spawn simultaneously
• Both return results
• Combined results make sense together

Scoring

• All 3 parts pass: You've mastered sub-agents!
• 2 parts pass: Review the failing pattern
• 1 part pass: Revisit spawning basics
• 0 parts pass: Start from "When to Split Agents"

Module Summary

You've learned:

• ✅ When to spawn sub-agents (and when not to)
• ✅ Coordination patterns (fire-forget, request-response, pipeline, parallel)
• ✅ How to build a researcher agent
• ✅ How to build an executor agent
• ✅ Parallel execution for speed

Key Insight: Sub-agents are for delegation, not everything. Spawn when the task is long, parallelizable, or needs isolation.

Next: Tools, APIs, and automation.