anchor.md is a lightweight, local-first AI context system designed to eliminate context loss when switching between LLMs or IDEs.

It operates on a strict local-first principle—all context files live securely within your repository. This means zero SaaS fees, zero cloud sync dependencies, and absolute data privacy. Because it uses standard Markdown, there's no vendor lock-in, and everything integrates perfectly with your existing Git workflow.

⚡ Key Features

• 🧠 Zero Context Loss: Seamlessly switch between Claude, ChatGPT, Gemini, Copilot, and others mid-task without repeating instructions.

• 🔌 Model-Agnostic: Compatible with any AI tool that supports local file reading.

• 🔒 Local-First & Secure: Your data stays in your repo. No cloud sync, no subscriptions, maximum privacy.

• 🌳 Git Native: Managed entirely via plain Markdown files. Read, edit, and version control your context just like your code.

• 🚦 Multi-Agent Safe: Strict synchronization rules prevent race conditions in multi-agent environments.

• 🎭 Role-Based Prompts: 6 pre-built expert roles (Backend, Frontend, DevOps, QA, Designer, Security) for hyper-precise AI interactions.

• ⚔️ Conflict Free: Clear, structured boundaries prevent different AI agents from stepping on each other's toes.

🚀 Getting Started

Bootstrap your project with a single command: npx create-anchor-md

Once installed, populate the @main.md file with your project details, tech stack, and architectural goals. Tag AI agents with @main.md or specific role prompts in your favorite IDE to give them instant, full-project context. The AI will then auto-sync its progress with files like TODO.md and CONTEXT_MEMORY.md.