Welcome to The Context Course

The Context Course is a free, hands-on course on doing machine learning with agent context. You'll learn to train models, optimize inference, and build datasets — all by defining agent context with SKILLS.md, Plugins, MCP, Subagents, and Hooks.

Whether you use Claude Code, Codex, or OpenCode, this course will teach you to give agents the right knowledge, tools, and structure to do real ML work.

How to join the course

First, you need to join this org to get info about the course:

Our first livestream will be May 12th, so sign up on YouTube.

🔴 Launch Live Stream & AMA

Weekly live AMA sessions covering course material, Q&A, and deep dives into ML with context engineering. First session: May 12th.

Deep Dive Projects

The course includes three hands-on ML projects that you'll tackle using agent context:

• Pre-training — Run and monitor pre-training experiments with agent-driven research workflows.
• Post-training — Fine-tune and evaluate chat models using a compact nano benchmark.
• Inference — Optimize local inference with llama.cpp, getting a GGUF model to run as fast as possible.

Each project is designed to be completed with a code agent — you'll define the context that makes it work.

Interactive Quizzes

Each unit includes quizzes to test your understanding. Complete them to track your progress and earn your certificate.

Supported Agents

This course works with multiple code agents. Instructions are provided for all three:

• Pi: Lean and Open Source agent.
• Claude Code: Anthropic's code agent (CLI & desktop)
• Codex: OpenAI's multi-agent CLI
• OpenCode: Open source agent from opencode.ai

If you use Cursor or GitHub Copilot, the concepts still transfer — their MCP and extension UX just isn't covered step-by-step.

Prerequisites

• Basic familiarity with Python and the command line
• A Hugging Face account
• At least one code agent installed (Claude Code, Codex, or OpenCode)

Contribute

This course is open source and contributions are welcome!

• Found a bug or want to improve content? Open an issue or PR on GitHub.
• Want to add a full section or unit? Open an issue first to discuss with the team.

Questions?

Open a community discussion on the org page, or ask during the weekly live AMA.