Cursor

Cursor speaks MCP natively. Add Mneva once and your agent has it in every chat in every workspace.

Where the config lives

Cursor reads MCP servers from ~/.cursor/mcp.json (global) or .cursor/mcp.json at the project root.

The paste

{
  "mcpServers": {
    "mneva": {
      "command": "npx",
      "args": ["-y", "mneva-connect"],
      "env": {
        "MNEVA_KEY": "mnv_your_key_here"
      }
    }
  }
}

Replace the key with one from mneva.dev/signup.

Restart Cursor. Open settings → MCP and you should see mneva with 16 tools listed. The first invocation pulls the npm package; subsequent runs hit the local cache.

Per-project vs global

• ~/.cursor/mcp.json — same brain in every workspace. If you want "Mneva, the partner who knows me everywhere" — this.
• .cursor/mcp.json in a repo — brain scoped to this project. Different repos can use different keys; useful when you're a contractor with multiple client codebases that shouldn't share memory.

You can mix: a global config with one brain plus a project-level config that overrides for a specific repo. Project config wins on collision.

For team usage with multiple Cursor users sharing one brain, see Multi-Agent — add MNEVA_AGENT per user so observations attribute correctly.