MCP memory-libsql-go

🧠 High-performance persistent memory system for Model Context Protocol (MCP) powered by libSQL. Features vector search, semantic knowledge storage, and efficient relationship management - perfect for AI agents and knowledge graph applications.

GitHub Stars

Language

6 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "zanzythebar-mcp-memory-libsql-go": {
      "command": "./mcp-memory-libsql-go",
      "args": [
        "-libsql-url",
        "file:./libsql.db"
      ],
      "env": {
        "LIBSQL_URL": "file:./libsql.db",
        "EMBEDDING_DIMS": "768",
        "LIBSQL_AUTH_TOKEN": "your-token",
        "METRICS_PROMETHEUS": "true",
        "EMBEDDINGS_PROVIDER": "openai"
      }
    }
  }
}

You run a high-performance, persistent MCP Memory Server backed by libSQL that stores model context data, supports vector search, and exposes MCP tools over stdio or SSE transports. This server is designed for fast embeddings-based retrieval, flexible multi-project management, and easy embedding model switching without migrating data.

How to use

Start the local server in single-database mode to serve MCP requests over stdio or enable SSE to accept MCP clients over HTTP. Use the default local database or specify a custom libsql database URL. You can also run in remote-libSQL or multi-project configurations as your needs grow. The server supports both text and vector search, plus optional hybrid search combining semantic and traditional text ranking.

How to install

Prerequisites: you need Go installed to build the server binary, and you should have libSQL CGO dependencies available on your system.

Install the binary to a standard location on your system:

make install

Minimal local build and run flow once installed:

./mcp-memory-libsql-go

To use a specific local database file, pass the libSQL URL as a flag:

./mcp-memory-libsql-go -libsql-url file:./my-memory.db

To connect to a remote libSQL instance, set LIBSQL_URL and LIBSQL_AUTH_TOKEN in the environment and start the server:

LIBSQL_URL=libsql://your-db.turso.io \
LIBSQL_AUTH_TOKEN=your-token \
./mcp-memory-libsql-go

If you want to expose the MCP server via SSE (HTTP), enable the SSE transport and specify the address and endpoint:

./mcp-memory-libsql-go -transport sse -addr :8080 -sse-endpoint /sse

Available tools

create_entities

Create new entities with observations and optional embeddings

search_nodes

Text or vector search for entities and relations

read_graph

Get recent entities and their relations

create_relations

Create relations between entities

delete_entity

Delete an entity and all its data

delete_relation

Delete a specific relation by source/target/type

add_observations

Append observations to an existing entity

open_nodes

Retrieve entities by names with optional relations

delete_entities

Bulk delete entities by name

delete_observations

Delete observations by id or content or all for an entity

delete_relations

Bulk delete relations

update_entities

Update entity metadata, embeddings, and observations

update_relations

Update relation tuples

health_check

Return server health and configuration

neighbors

1-hop neighbors for given entities

walk

Graph expansion by bounded-depth walk

shortest_path

Shortest path between two entities

Built by

VeilStrat

AI signals for GTM teams