theDakshJaitly

theDakshJaitly / mex

Public

Persistent project memory for AI coding agents. Structured scaffold + drift detection CLI.

www.launchx.pagemex claude-code claude-code-skills cli-tool context-management cursor
17
1
100% credibility
Found Mar 28, 2026 at 17 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Shell
AI Summary

mex creates a navigable markdown structure and checks to keep AI coding assistants aware of project state, architecture, and patterns across sessions.

How It Works

1
🧠 Discover mex

You learn about mex, a helpful system that gives your AI coding buddy a lasting memory of your entire project so it never forgets what you built.

2
📥 Add to your project

You bring mex into your project folder with a simple download and placement.

3
✨ Run easy setup

You launch the setup which scans your work, asks which AI helper you use, and prepares everything in minutes.

4
🤖 AI builds memory guides

Your AI reads your project and creates clear guides on how things connect, rules to follow, and steps for common jobs.

5
🔍 AI uses smart navigation

Now your AI starts each chat by checking a quick map to load only the needed memory, keeping chats fast and focused.

6
✅ Spot and fix drifts

You check anytime to see if memory matches your latest changes, and easily update just what's off.

🎉 Perfect project memory

Your AI always remembers your project's details perfectly, making work smoother and mistake-free forever.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 17 to 17 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is mex?

mex is a persistent project memory system for AI coding agents, using a structured markdown scaffold in a .mex/ folder to store architecture, conventions, decisions, and task patterns. It solves the cold-start problem where agents forget prior work, bloat context windows with mega-CLAUDE.md files, or drift from codebase reality—via a Shell/TS CLI that scans for inconsistencies like stale paths or missing deps, then generates targeted AI prompts to fix them. Setup clones to .mex/, runs bash setup.sh for initial population, and supports Claude Code, Cursor, Windsurf, or Copilot via root config files.

Why is it gaining traction?

Unlike ad-hoc docs or github persistent storage hacks, mex auto-grows from real tasks, enforces edges between files for navigable context (under 150 tokens bootstrap), and runs zero-token drift checks with mex check—flagging dead edges or undocumented scripts before AI hallucinates. Post-commit hooks silently score changes, and mex sync loops check-fix-verify, making github mex a low-friction way to keep agents honest across JS/Python/Go stacks. The visualize.sh graph hooks devs visually.

Who should use this?

AI-heavy teams using Claude Code or Cursor on mid-sized repos (e.g., APIs, pipelines) where session amnesia kills velocity—think backend devs maintaining Twilio/Supabase flows or fullstackers debugging multi-provider STT. Skip for toy scripts or non-AI workflows.

Verdict

Grab it if you're all-in on agentic coding; the CLI (mex check/sync/watch) delivers immediate wins despite 17 stars and 1.0% credibility score signaling early days. Docs shine, but watch for edge cases in non-bash envs—solid foundation for persistent project memory. (198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.