Course Wiki

Context Engineering

Feeding coding agents durable project knowledge through AGENTS.md so they stop re-learning the codebase every session.

An agent that starts every session knowing nothing about your project will fight it. Context engineering is the module’s practice of writing the durable knowledge down: AGENTS.md at the repo root carries the project’s structure, conventions, commands, and constraints; task-level context rides in the backlog items; the agent reads both before touching code.

The course treats the instruction file as living documentation — updated as the project evolves, kept short enough to stay in the agent’s working set. The payoff compounds across modules: subagents, skills, and CI all assume the shared context file is accurate.

Taught in

• LLM Zoomcamp

  • Module 3: Orchestration
    • AI Orchestration
    • Context Engineering
    • Next Steps

• AI Dev Tools Zoomcamp

  • Module 1: AI-Native Developer Workflow
    • AI-Native Developer Workflow
  • Module 2: Build and Ship an AI-Assisted Full-Stack App
    • Build and Ship an AI-Assisted Full-Stack App
  • Module 5: Coding Agent Building Blocks: Reusable Skills and Specialized Subagents
    • Module 5 — Coding Agent Capabilities: MCP, Skills, Plugins, and Custom Agents

• Data Engineering Zoomcamp

  • Module 2: Workflow Orchestration
    • 2.5.2 - Context Engineering with ChatGPT ## Related concepts

• Coding Agents

• Spec-Driven Development

• Agent Skills and Subagents

• Loop and Graph Engineering