Connect MyMemory to Cursor
Updated
Definition and quick answer
MyMemory for Cursor is a Model Context Protocol (MCP) connection that lets Cursor's AI agent search, save and read your MyMemory library while you code.
How to connect: Click Add to Cursor from Add memory → MCP in the app for a one-click install, run npx @everstash/mcp setup, or copy the MyMemory MCP config and add it to ~/.cursor/mcp.json with your sm_ API key, then reload MCP servers in Cursor. The agent gets five tools: save_memory, search_memories, list_spaces, get_memory_by_id and delete_memory.
- Connection type
- Model Context Protocol (MCP), stdio transport
- Supported clients
- Cursor
- Authentication
- MyMemory API key (sm_…)
- Requirements
- Node.js 18+, a MyMemory API key
- Package
- @everstash/mcp on npm (runs with npx)
- Tools exposed
- 5
Setup steps
1. Add to Cursor with one click
Open Add memory → MCP in the app and click Add to Cursor. Cursor opens with the server pre-filled; paste in your API key when Cursor asks for it.
2. Or run the setup wizard
npx @everstash/mcp setup finds the MCP clients installed on your computer, asks which to connect and for your API key, then writes each one's config itself, backups included.
npx @everstash/mcp setup3. Create an API key
Generate a key on the MyMemory dashboard (the API key card on the Overview tab). Keys start with sm_ and grant access to your account, so keep the config file private.
4. Open the MCP tab in MyMemory
Sign in at mymemory.dev, click Add memory, and open the MCP tab. Pick your client to see its exact configuration block with a copy button, or use the one-click install button where available.
5. Or add the server manually
Paste the block from the MCP tab into ~/.cursor/mcp.json (all projects) or .cursor/mcp.json (one project) with your API key, then reload MCP servers in Cursor settings.
{ "mcpServers": { "everstash": { "command": "npx", "args": [ "-y", "@everstash/mcp" ], "env": { "EVERSTASH_API_KEY": "sm_your_api_key_here" } } } }
What Cursor can do with MyMemory
| Tool | What it does |
|---|---|
save_memory | Saves text, a note or a URL to your library. |
search_memories | Runs hybrid semantic and full-text search; returns up to 100 results. |
list_spaces | Lists the spaces your account can access. |
get_memory_by_id | Returns the full content of one memory. |
delete_memory | Previews a memory, then deletes it only when called with confirm: true. |
Limitations
- The local server (npx @everstash/mcp) runs on your computer over stdio, so each machine needs Node.js 18+ and its own config. For clients that add a remote MCP server by URL instead, point them at https://api.mymemory.dev/v1/mcp and sign in with your MyMemory account (or send your API key as a Bearer token where the client supports headers) — nothing to install.