Patdolitse

Patdolitse / engram

Public

Local AI memory layer for Claude Code, Codex, Cursor, and other MCP-compatible tools.

github.comPatdolitseengram ai-agent ai-memory ai-tools claude-code codex
12
0
100% credibility
Found May 19, 2026 at 34 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

Engram is a local memory layer for AI coding tools that solves the problem of constantly re-explaining yourself when switching between different AI assistants or starting new sessions. It stores your identity, work preferences, quality standards, lessons learned, and key decisions as simple files on your computer. These files can be read by any compatible AI tool, so Claude Code, Codex, Cursor, and others all start from the same understanding of who you are and how you like to work. The project is open source, stores everything locally on your machine, and includes tools for backing up and migrating your data.

How It Works

1
💻 You use an AI coding assistant

You start using tools like Claude Code, Codex, or Cursor to help you build projects and solve problems.

2
😓 You keep repeating yourself

Every time you switch tools or start a new session, you have to explain who you are, how you like to work, and what you've learned before.

3
🧠 You discover Engram

Engram is a personal memory layer that remembers everything about how you work, so your AI tools can understand you instantly.

4
📦 You set up your memory in minutes

A simple setup walks you through creating your identity card, work preferences, and quality standards—all stored safely on your own computer.

5
You connect Engram to your AI tools
🔗
Direct connection

For tools like Claude Code and Codex, you enable a built-in connection so they automatically read your memory at the start of every session.

📝
Identity card

For other AI tools, you copy a simple text card that tells the AI who you are and how you like to work.

6
✨ Your AI tools start remembering

As you work, you can save lessons learned, key decisions, and project context. Your AI tools pull this knowledge automatically.

🎉 Every AI tool knows you instantly

Switching tools or starting new sessions is seamless—your AI assistant immediately understands your style, preferences, and past experiences.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 34 to 12 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 engram?

Engram is a local memory layer for AI coding tools. Think of it as giving your AI assistant a persistent memory that works across Claude Code, Codex, Cursor, and any MCP-compatible tool. All your context, preferences, lessons learned, and key decisions live in plain JSON and Markdown files under `~/.engram/`, exposed through the Model Context Protocol so multiple tools can read the same information. Instead of re-explaining yourself every session, the AI just loads your identity card and knows who you are.

Why is it gaining traction?

The hook is portability and tool-agnosticism. Cloud memory solutions lock you into one platform; Engram keeps your data in files you own and can edit directly. The MCP integration means it slots into your existing workflow without vendor lock-in. You can export your entire memory to OpenClaw format, back it up as JSON, or migrate it to another machine. For developers running multiple AI tools in parallel, this solves the context-fragmentation problem that usually forces you to start fresh with each tool.

Who should use this?

Developers who use Claude Code or Codex alongside other AI assistants and are tired of repeating themselves. Product managers working with AI tools who want to preserve project decisions without managing a separate wiki. Anyone who values data ownership over convenience. If you switch tools frequently or work across projects, Engram prevents the "new session, zero context" problem.

Verdict

Engram solves a real pain point with a clean architecture, but with only 12 stars and a 1.0% credibility score, it is very early-stage. The Python codebase and MCP compliance look solid based on the README, but test coverage and documentation are minimal. Use it if you want to experiment with cross-tool memory; wait for more community validation if you need production stability.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.