Embedding Search

Provides vector-based search over transcript embeddings stored in Turso with metadata like episode, timestamps, and similarity scores.

javascript

GitHub Stars

javascript

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": {
    "spences10-mcp-embedding-search": {
      "command": "node",
      "args": [
        "/path/to/mcp-embedding-search/dist/index.js"
      ],
      "env": {
        "TURSO_URL": "https://turso.example.com/db",
        "TURSO_AUTH_TOKEN": "YOUR_TURSO_TOKEN"
      }
    }
  }
}

You can use the mcp-embedding-search server to query a Turso database of embeddings and transcript segments. It returns the most relevant transcript snippets based on vector similarity, along with episode metadata and timings, helping you locate the right parts of a show or podcast quickly without creating new embeddings.

How to use

You interact with this MCP server through your MCP client by calling the search_embeddings tool. Provide your question as the query text, and you can tune which results you want by setting a limit and a minimum similarity score. The server returns transcript segments that best match your question, including the episode title, the segment text, and the start and end times for precise navigation.

How to install

Prerequisites: you need Node.js installed on your system. You also need access details for your Turso database (URL and access token). You will run the server locally in development mode first, then configure your MCP client to connect to it.

# Install dependencies
npm install

# Build the project
npm run build

# Run in development mode
npm run dev

Configuration and usage notes

Configure your MCP client to run the embedding search server as a local process. Use the following configuration snippet to start the server via a local runtime. Replace the placeholder path with the actual build output location on your system, and set your Turso connection details.

{
  "mcpServers": {
    "mcp_embedding_search": {
      "command": "node",
      "args": ["/path/to/mcp-embedding-search/dist/index.js"],
      "env": {
        "TURSO_URL": "your-turso-database-url",
        "TURSO_AUTH_TOKEN": "your-turso-auth-token"
      }
    }
  }
}

API usage at a glance

Once started, you can invoke the search_embeddings tool from your MCP client by supplying a question, and optionally a limit and a minimum similarity score. The server returns a list of matching transcript segments with their episode title, segment text, start and end times, and the cosine similarity score.

Notes on security and performance

The server uses a Turso database connection pool for efficient queries and includes error handling to gracefully manage issues such as connectivity problems or invalid input. Ensure your Turso credentials are kept secure and rotated as needed.

Troubleshooting

If the server fails to start, verify that Node.js is installed, the build artifacts exist at the expected path, and the Turso URL and token are correct. Check your MCP client configuration to ensure the server process is launched with the proper environment variables.

Available tools

search_embeddings

Searches for relevant transcript segments using vector similarity. Accepts a question, limit, and minimum score; returns a list of matching segments with metadata and similarity scores.

Built by

VeilStrat

AI signals for GTM teams