CursorPool
← 返回首页

Memstate AI

Versioned, structured memory for AI agents. Every fact is tracked, every change is logged, and your agent always gets the current answer — not a pile of outdated context. In complex multi-session coding tasks, Memstate AI outperforms mem0, 84.4% vs 20.4%, for memory search and recall.

cursor.directory·25
MCP

Memstate AI

Versioned, structured memory for AI agents. Store, retrieve, and version factual memory with automatic conflict detection.

{"command": "npx", "args": ["-y", "@memstate/mcp"], "env": {"MEMSTATE_API_KEY": "YOUR_API_KEY_HERE"}}
规则

use-memstate-memory-system.md

Instructs AI agent how to fetch or save task memories.

# Memstate Memory

This project uses the Memstate MCP server for persistent, versioned memory.

## Before each task
Call `memstate_get(project_id="<your_project>")` to load existing context.

## After each task
Call `memstate_remember(project_id="<your_project>", content="<markdown summary>")` to save what was done.

## Project naming
Use a short snake_case name matching your repo or topic (e.g. `my_app`, `api_service`). Keep all related memories in the same project.

## Tools at a glance
- `memstate_remember` — save markdown summaries, notes, architecture docs
- `memstate_get` — retrieve all memories for a project
- `memstate_search` — semantic search across memories
- `memstate_set` — store a single key/value fact
- `memstate_history` — view version history of a memory

来源:https://github.com/memstate-ai/memstate-mcp