MCP.so
登录

Yourmemory MCP Server

@sachitrafa

基本信息

分类

其他

许可证

NOASSERTION

传输方式

stdio

作者

sachitrafa

提交者

sachit mishra

关于 Yourmemory MCP Server

Agentic AI memory with Ebbinghaus forgetting curve decay. +16pp better recall than Mem0 on LoCoMo.

Yourmemory MCP Server 配置

暂无标准配置

该服务器的 README 中没有可解析的 MCP 配置块,请前往代码仓库查看安装说明。

代码仓库

Yourmemory MCP Server 工具

工具列表暂未提供

当前收录暂时没有可展示的工具列表,不代表服务没有工具。请查阅项目文档了解支持的能力。

工具是从 README 中自动提取的。维护者可以在 ## Tools 标题下列出工具,即可填充这部分内容。

概览

What is Yourmemory?

Yourmemory provides persistent, self-improving memory for AI agents, built on the science of how humans remember. It integrates with any MCP client (Claude, Cursor, Cline, Windsurf) and runs entirely on your machine with no API key required.

How to use Yourmemory?

Install with pip install yourmemory, then run yourmemory-register <your-token> and yourmemory-setup. Optionally download a standalone binary from the latest release. Start the server with no arguments. Works out of the box with DuckDB (default) or Postgres with pgvector for teams.

Key features of Yourmemory

  • Consolidation: compresses related facts into clean summaries.
  • Biological decay: memories fade on the Ebbinghaus forgetting curve.
  • Entity graph: memories link by shared people, places, and concepts.
  • Survives context resets: hands back working context after compaction.
  • Tamper-evident audit trail: hash-chained ledger logs every operation.
  • Team memory pools: role-based shared and private memory.
  • Data rights: export and right‑to‑forget with one command.
  • MCP‑native and local‑first: no external API or cloud dependency.

Use cases of Yourmemory

  • AI agents remembering user preferences and context across sessions.
  • Teams sharing institutional knowledge with role‑based access.
  • Long‑running tasks that need to survive context window resets.
  • Compliance‑heavy environments requiring an auditable memory trail.
  • Research and experimentation with memory consolidation and decay.

常见问题

1

What makes Yourmemory different from other memory tools?

It uses consolidation, biological decay, an entity graph, and hybrid retrieval — not just a vector database. Memory gets sharper over time and recalls what you forgot to ask for.

2

What are the runtime requirements?

Python 3.11–3.14. No Docker or database setup is needed. All memory is stored locally in `~/.yourmemory/` by default (DuckDB). Optionally install Ollama for better local fact extraction.

3

Where does my data live?

All data is stored locally on your machine. DuckDB is the default backend; you can switch to Postgres with the pgvector extension for shared or production use.

4

How does memory decay work?

Each memory ages on an Ebbinghaus forgetting curve. Stale, unused facts fade, while frequently recalled ones persist. Different categories (strategy, fact, assumption, failure) have different half‑lives.

5

Is there an audit trail?

Yes. Every read, write, update, delete, and consolidation is appended to a hash‑chained, tamper‑evident audit log. Altering one record breaks the chain.

评论

其他 分类下的更多 MCP 服务器