context-methodology

The context-methodology uses git worktrees to maintain two independent branches: a `context` branch for CLAUDE.md and .claude/ directory, and a `main`/`master` branch for project code. This separation keeps AI prompts isolated from application code history while both live in the same repository.

The context branch methodology separates AI assistant configuration from source code using git worktrees and dual branch histories. Use it when you store AI instructions in CLAUDE.md, need isolated AI contexts for experiments, and want to manage feature development and AI configuration updates independently.

Create feature branches from the default branch, then use git worktrees to establish isolated working directories. The context methodology guides you through hooking up AI config and managing pull requests while maintaining strict separation between code and AI configuration histories.

Yes. The context branch stores both CLAUDE.md and the .claude/ directory on a dedicated `context` branch, keeping all AI configuration separate from your main codebase while supporting automated detection of the default branch and PR creation workflows.

Separating AI configuration keeps your codebase clean, prevents AI instruction changes from polluting code history, enables independent experimentation on AI contexts, and maintains dual independent commit histories optimized for their respective concerns.

The context methodology provides a guided workflow: maintain the context branch for CLAUDE.md and .claude/ updates, create feature branches from the default branch for code work, use worktrees for isolation, and manage pull requests without mixing AI and code histories.