HKUDS

HKUDS / MGP

Public

Memory Governance Protocol

18
1
100% credibility
Found Apr 15, 2026 at 18 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

MGP is an open protocol with reference tools for AI systems to standardize secure, governed persistent memory storage, recall, and auditing across various backends.

How It Works

1
🔍 Discover safe AI memory

While building an AI helper that needs to remember user likes and facts securely, you find MGP, a simple way to organize and protect those memories.

2
📥 Get the memory manager

Download the free tool and start it on your computer with one easy command, like flipping a switch.

3
🔗 Pick your memory notebook

Choose where to keep memories, like a simple folder on your desk or a secure notebook app—it connects automatically.

4
💾 Save a user preference

Tell it to remember something personal, like 'user loves dark mode', and it stores it safely.

5
🔎 Recall what was saved

Ask for memories about 'dark mode', and it brings back exactly the right ones, ready to use.

6
🧹 Clean up old memories

Update, expire, or erase memories as life changes, keeping everything fresh and private.

✅ AI remembers perfectly

Your AI helper now recalls user details reliably and safely, no matter where memories are stored, making chats smarter and more personal.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 18 to 18 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 MGP?

MGP is a Python-based open protocol that standardizes governed memory for AI systems, letting agent runtimes write, search, update, expire, and audit persistent memories across backends like Postgres, LanceDB, or in-memory stores. It solves the chaos of heterogeneous memory managers by providing a single API layer—your agent issues one request, and any compliant backend handles it with built-in policy context and audit trails. Users get a reference FastAPI gateway, Python SDK, and quick-start Docker setup to spin up a memory service in minutes.

Why is it gaining traction?

It stands out with pluggable adapters for seven backends out of the box, full lifecycle ops like revoke and purge, and machine-verifiable compliance tests that prove interoperability. Developers hook into it fast via curl or SDK for write/search flows, plus it's a peer to MCP for tools—use MGP for memory governance without rewriting agent logic. The schema-driven OpenAPI spec and audit query endpoint make it dead simple to integrate governed recall into production AI pipelines.

Who should use this?

AI runtime builders adding persistent memory to agents, like those extending LangGraph or custom LLM chains needing user prefs or facts stored securely. Platform engineers deploying multi-tenant memory backends who want governance without vendor lock-in. Teams evaluating memory github copilot alternatives or github memory manager solutions for scalable, auditable AI state.

Verdict

Grab it for early prototypes or MCP-paired stacks—solid docs, Makefile-driven tests, and v0.1.0 protocol make it playable now, despite 19 stars and 1.0% credibility score signaling pre-mainstream maturity. Skip for battle-tested prod unless you're adapter-hacking.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.