
Build a multi-agent coding script
You'll have a working Python or Node script that routes coding tasks to specialized AI agents (code review, testing, documentation) and runs them in sequence on your actual codebase.
What you'll make
Multi-Agent Coding Script
- agent_definitions.py — the specialized agent personas and prompts: what to include
- task_router.py — coordination logic for routing tasks to appropriate agents: what to include
- main.py — orchestration that runs agents in sequence and handles handoffs: what to include
- config.json — settings for your codebase paths and agent parameters: what to include
Paste your complete Python script files, ensuring the main.py can be run from command line on a real code file
The 3 parts
Each part ends with a check you can see.
Design Agent Personas
Create three specialized AI agent personas for code review, testing, and documentation, each with focused prompts that address distinct failure modes
Done when
Three agent persona definitions with specific prompts, clear input/output formats, and documented failure modes each agent should catch
Build Task Router
Create coordination logic that routes code files to the right agents in sequence and handles the handoffs between them
Done when
Working task_router.py that takes a code file, runs it through three agents in order, and passes each agent's output to the next without breaking
Test and Calibrate
Run the complete script on real code from your projects and calibrate each agent based on what they get right vs wrong
Done when
Complete main.py script tested on at least 3 real code files with documented calibration notes for when to trust each agent's output
Capstone: you put together your Multi-Agent Coding Script.
Where this helps
An elective: a single lesson that sits outside rtwo's job certificates and doesn't count toward one.
About this lesson
What you'll be able to do
- Run a working multi-agent script that routes coding tasks to specialized AI agents and executes them in sequence
- Identify which types of code review, testing, and documentation tasks work well with AI agents vs where you need human oversight
- Have a tested workflow for coordinating multiple AI agents on your actual codebase without them stepping on each other
Who it's for
- Developers who want to experiment with AI agent coordination on real projects
- Technical leads curious about delegating routine coding tasks to AI systems
- Anyone with basic Python/Node experience who wants to see what 'agentic coding' actually looks like in practice
What it won't cover
- Building your own LLM from scratch or training custom models
- Enterprise-scale agent orchestration with complex authentication
- Replacing your entire development workflow — this is about augmenting specific tasks
Skills you'll practise
- Multi-agent task coordination
- AI output validation and filtering
- Agent persona design for coding contexts
- Iterative workflow debugging