Summary

mex provides persistent, navigable project memory for AI coding agents, solving the problem of session-based forgetting and context window overload. It targets developers using AI agents, offering a structured markdown scaffold and a CLI to keep the agent's understanding synchronized with the codebase, reducing token costs and improving focus.

How It Works

mex uses a structured markdown scaffold (.mex/) for persistent project knowledge (architecture, conventions, patterns). A CLI performs drift detection by validating this scaffold against the live codebase using eight specialized checkers, ensuring accuracy without excessive token use. After each task, the agent updates the scaffold, automatically creating new patterns from real work, allowing project memory to compound over time.

Quick Start & Requirements

Installation involves cloning the repository and running bash .mex/setup.sh, which auto-builds the CLI, scans the codebase, and populates the scaffold in ~5 minutes. A bash-compatible shell (WSL, Git Bash) is required for shell scripts; Node.js is needed for direct CLI execution. The project is licensed under MIT.

Highlighted Details

• Drift Detection: Eight checkers validate scaffold integrity against the codebase without AI.
• CLI Functionality: Commands include mex check (drift scoring), mex sync (AI-driven correction), mex init (AI brief generation), and mex watch (post-commit hook).
• Scaffold Growth: The scaffold automatically expands by creating new patterns from agent tasks.
• Multi-Tool Compatibility: Configuration files are interchangeable across AI coding tools like Claude Code, Cursor, and Copilot.

Maintenance & Community

The project includes CI workflow badges. No specific details regarding maintainers, community channels, or sponsorships are present in the provided README.

Licensing & Compatibility

Released under the MIT license, which is permissive for commercial use and integration into closed-source projects. Designed to work with any technology stack (JavaScript, Python, Go, Rust).

Limitations & Caveats

Core shell scripts require a bash-compatible environment. While Node.js CLI commands offer broader platform compatibility, reliance on bash for setup may be a consideration. Effectiveness depends on the AI agent's ability to accurately update the scaffold.