MCP Server MCP

Local Memory

By github 👁 1 views ▲ 0 votes

Persistent local AI memory. SQLite + FTS5 + vector search, knowledge graph. No cloud, no API keys.

Homepage Verified source GitHub

Install

npx -y @studiomeyer/[email protected]

Configuration Example

{
  "remotes": [],
  "packages": [
    {
      "registryType": "npm",
      "identifier": "@studiomeyer/local-memory-mcp",
      "version": "2.4.3",
      "transport": {
        "type": "stdio"
      },
      "environmentVariables": [
        {
          "description": "Custom path for the SQLite database file",
          "format": "string",
          "name": "MEMORY_DB_PATH"
        },
        {
          "description": "Disable the local embedding pipeline (forces FTS5-only search). Useful in restricted networks where the model can't be downloaded.",
          "format": "string",
          "name": "MEMORY_EMBED_DISABLED"
        },
        {
          "description": "Override the embedding model id (defaults to Xenova/multilingual-e5-small).",
          "format": "string",
          "name": "MEMORY_EMBED_MODEL"
        },
        {
          "description": "Override the Transformers.js cache directory (defaults to the OS data dir).",
          "format": "string",
          "name": "MEMORY_EMBED_CACHE_DIR"
        }
      ]
    }
  ]
}

mcp model-context-protocol npm