context-engineering-intro

Context Engineering Intro is a template repository for getting more reliable results from AI coding assistants. The central argument is that most failures when using AI to write code are not model failures but context failures: the AI does not have enough information about what you are building, how your project is structured, or what patterns you follow. This template gives you a structured starting point for supplying that context.

The main pieces are a CLAUDE.md file containing project-wide rules that the AI reads at the start of every session, an INITIAL.md file where you describe a feature you want to build, and an examples folder where you place existing code the AI can learn patterns from. The README treats the examples folder as critical: concrete code samples of your file structure, testing conventions, and integration patterns produce better outputs than text descriptions alone.

The workflow has two steps handled by custom slash commands. Running /generate-prp reads your feature description, analyzes the codebase for existing patterns, gathers relevant documentation, and writes a detailed implementation plan called a PRP (Product Requirements Prompt). Running /execute-prp reads that plan and carries out the work step by step, running tests and fixing issues as it goes.

The repository ships with the command definitions for both slash commands, a base PRP template, an example feature request, and a completed example PRP. You use it by cloning it, editing the rules and description files for your own project, and building out the examples folder. The README notes the workflow is centered on Claude Code but the approach applies to other AI coding assistants as well.