What is MemPalace and how does it give an AI memory?

MemPalace is a free, open-source, local-first AI memory system that stores your conversation and project history as verbatim text and retrieves it with semantic search. It creates a structured memory layer outside the model so your AI assistant can recall exact past context instead of starting every chat from zero.

How do you install MemPalace?

MemPalace is a Python tool that installs via uv (uv tool install mempalace) or pip (pip install mempalace). After installing, run mempalace init ~/projects/myapp to initialize it for your project.

How does MemPalace compare to Mem0 on recall accuracy?

On the LongMemEval benchmark, MemPalace reports a 96.6% recall rate versus 89.2% for Mem0. MemPalace also runs fully locally with zero API calls, while Mem0 requires a paid OpenAI API.

How does MemPalace organize stored memory?

MemPalace uses a palace metaphor with three levels: Wings for people and projects, Rooms for topics within those projects, and Drawers that hold the original content stored verbatim. This lets you scope a search to a specific project wing or topic room instead of querying a flat vector database.

Does MemPalace work with Claude Code and other AI tools?

Yes. MemPalace ships native plugins including a .claude-plugin directory for Claude Code, a .codex-plugin directory for OpenAI Codex, and a .agents/plugins directory for MCP-compatible tools, plus support for Gemini CLI and local models. It exposes an MCP-compatible endpoint out of the box for persistent coding-agent memory.

MemPalace vs Mem0：96.6% 回忆基准 & 最佳 AI 记忆框架（2026）

MemPalace is the best-benchmarked open-source AI memory system. Learn how to install it, mine your project history, and retrieve context with semantic search so your AI assistant never forgets again.

Go
Python
MIT
更新于 2026-05-15

{</* 资源信息 */>}

每一次 AI 对话都是从零开始的。你需要一遍又一遍地解释你的技术栈、你的偏好和你的项目历史。MemPalace 解决了这个问题。它是经过最佳基准测试的开源 AI 记忆系统，并且完全免费。

拥有 51,745 个 GitHub 星标 以及不断增长的插件生态系统，MemPalace 将你的对话历史以原文形式存储，并通过语义搜索进行检索。你的 AI 助手终于能记住了。

📦 出现在以下合集中

📚 知识库 Stack →

🔗 相关资源推荐

📦 出现在以下合集中

💬 留言讨论