Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Memory Palace is a long-term memory operating system purpose-built for AI Agents.
| Date | Stars |
|---|---|
| 2026-07-31 | 309 |
| 2026-08-06 | 309 |
Today
— stars today
This week
— stars this week
This month
— stars this month
Momentum
0.0
growth rate 0.00%/day
<p align="center"> <img src="docs/images/系统架构图.png" width="280" alt="Memory Palace Logo" /> </p> <h1 align="center">🏛️ Memory Palace</h1> <p align="center"> <strong>Persistent memory for AI agents — searchable, auditable, cross-session.</strong> </p> <p align="center"> <em>"Every conversation leaves a trace. Every trace becomes memory."</em> </p> <p align="center"> <img src="https://img.shields.io/badge/license-MIT-blue.svg" alt="License" /> <img src="https://img.shields.io/badge/python-3.10+-3776ab.svg?logo=python&logoColor=white" alt="Python" /> <img src="https://img.shields.io/badge/FastAPI-009688.svg?logo=fastapi&logoColor=white" alt="FastAPI" /> <img src="https://img.shields.io/badge/React-18-61dafb.svg?logo=react&logoColor=black" alt="React" /> <img src="https://img.shields.io/badge/Vite-646cff.svg?logo=vite&logoColor=white" alt="Vite" /> <img src="https://img.shields.io/badge/SQLite-003b57.svg?logo=sqlite&logoColor=white" alt="SQLite" /> <img src="https://img.shields.io/badge/protocol-MCP-orange.svg" alt="MCP" /> <img src="https://img.shields.io/badge/Docker-ready-2496ed.svg?logo=docker&logoColor=white" alt="Docker" /> </p> <p align="center"> <a href="README_CN.md">中文</a> · <a href="https://agi-is-going-to-arrive.github.io/Memory-Palace/">Landing Page</a> · <a href="docs/README_EN.md">Docs</a> · <a href="docs/GETTING_STARTED_EN.md">Quick Start</a> · <a href="docs/EVALUATION_EN.md">Benchmarks</a> </p> --- ## What Is Memory Palace? Memory Palace gives LLM agents a persistent, searchable, and auditable memory store, so each conversation can build on the last instead of starting from scratch. Through the [MCP (Model Context Protocol)](https://modelcontextprotocol.io/), one backend serves **Claude Code, Codex, Gemini CLI, and OpenCode** as full skill clients. For IDE hosts (`Cursor / Windsurf / VSCode-host / Antigravity`), use the repo-local **`AGENTS.md` + rendered MCP snippet** path instead. Shortest path: [SKILLS_QUICKSTART_EN.md](docs/skills/SKILLS_QUICKSTART_EN.md) for CLI clients; [IDE_HOSTS_EN.md](docs/skills/IDE_HOSTS_EN.md) for IDE hosts. If you want an AI to guide installation step by step, start from [`memory-palace-setup`](https://github.com/AGI-is-going-to-arrive/memory-palace-setup). The recommended stance is **skills + MCP**, not MCP-only. ### Why Memory Palace? | Pain Point | How Memory Palace Solves It | |---|---| | 🔄 Agent forgets between sessions | Persistent SQLite-backed memory that survives across runs | | 🔍 Hard to recall past context | Hybrid retrieval (keyword + semantic + reranker) with intent-aware routing | | 🚫 No control over what gets stored | Write Guard pre-checks every write; snapshots enable full rollback | | 🧩 Each client needs its own integration | One unified MCP interface | | 📊 Can't observe what's happening | Dashboard with Memory, Review, Maintenance, and Observability views | --- ## What's New <p align="center"> <img src="docs/images/memory_palace_upgrade.png" width="900" alt="Memory Palace Project Upgrade Comparison" /> </p> - **Memory Maintenance Engines** (v2): four independent engines — Forgetting (vitality decay + archive), Layering (L0→L1→L2 provenance), Compression (cascade preview), Procedural (step extraction). All read-only preview by default; mutations gated behind explicit review tokens. - **SSE Hardening**: per-principal rate limiting (429 + Retry-After), idle watchdog, trusted proxy CIDR allowlist, graceful shutdown draining, loopback port auto-fallback. - **Dashboard**: L0/L1/L2 layer hierarchy, forgetting simulation, archive candidates, Search Quality panel, Observability SSE live view with connection-loss banner. - **Retrieval**: RRF fusion remains on by default for Profile B only (`RRF_K=10`). Profile C/D keep RRF and MMR disabled by default and enable the sqlite-vec native vector engine when the pip `sqlite-vec` package is installed; otherwise it falls back to legacy scoring. Mixed CJK/Latin handling, full-width normal
Excerpt of 25,269 characters
Read on GitHubWould you bet a product on this? Bounded 0–100 and slow moving.
matched fp:ed2716d3a67787b8, llm:Description: 'Memory Palace is a long-term memory operating system purpose-built for AI Agents.' Language: Python. No topics provided.
matched fp:ed2716d3a67787b8, llm:Description: 'Memory Palace is a long-term memory operating system purpose-built for AI Agents.' Language: Python. No topics provided.
matched fp:ed2716d3a67787b8, llm:Description: 'Memory Palace is a long-term memory operating system purpose-built for AI Agents.' Language: Python. No topics provided.