Cole Medin's Context Engineering Method
Table of content
Cole Medin is the founder of Dynamous AI, an AI education platform, and CTO of Ottomator.ai. His YouTube channel has 185,000+ subscribers learning to build AI agents and use AI coding assistants effectively. He created Archon, an open-source AI agent that builds other AI agents.
Context Engineering
Medin’s core insight: AI coding assistants fail not because of the model, but because of the context you provide. He calls the solution “context engineering.”
The term describes a structured approach to giving AI coding assistants everything they need upfront: architecture decisions, best practices, project rules, code examples, and validation processes.
From his Context Engineering 101 video:
“Context engineering is the new vibe coding. It’s the way to actually make AI coding assistants work.”
The method uses PRPs (Product Requirements Prompts), detailed blueprints that specify exactly what the AI should build and how. Instead of iterating through broken code, you front-load the context.
| Component | Purpose |
|---|---|
| Architecture spec | How pieces connect |
| Code examples | Patterns to follow |
| Project rules | Conventions to enforce |
| Validation gates | How to verify success |
The Three-Layer Workflow
Medin teaches a process similar to spec-driven development:
- Plan with extensive context (PRPs, rules, examples)
- Implement using AI coding assistants
- Validate with defined gates
The key difference from casual prompting: you treat the prompt and context as engineered artifacts, not throwaway messages.
Archon
Archon is Medin’s open-source project for building AI agents that build other AI agents. It has 13,600+ GitHub stars.
The idea: instead of manually coding each AI agent, use an AI agent specialized in creating agents. Archon serves as a knowledge and task management backbone for AI coding assistants.
From his Archon launch video:
“Me and a couple other guys have been overhauling Archon to make it the knowledge and task management backbone for AI coding assistants.”
Teaching Philosophy
Medin’s approach favors practical application over theory. His videos walk through real builds, not toy examples. He runs the Dynamous Agentic Coding Course and teaches workshops at conferences like AI Coding Summit.
His Top 20 Lessons from Building 100s of AI Agents condenses years of iteration into actionable takeaways.
Technical Stack
Based on his GitHub repos and videos:
- AI Agents: Pydantic AI, LangGraph, n8n
- RAG systems: Custom vector stores, Vectorize
- Local AI: Self-hosted model deployments
- AI Coding: Claude Code, Cursor, Windsurf
- MCP Servers: Archon MCP, custom integrations
Key Takeaways
| Principle | Implementation |
|---|---|
| Engineer your context | Write PRPs before coding |
| Validate systematically | Define success criteria upfront |
| Build agents to build agents | Use Archon for agent creation |
| Reset context between phases | Separate planning from execution |
Links
- YouTube: Cole Medin
- GitHub: coleam00
- Dynamous AI
- Ottomator.ai
- Archon on GitHub
- Context Engineering Intro
Next: Swyx’s Learning in Public
Get updates
New guides, workflows, and AI patterns. No spam.
Thank you! You're on the list.