Atlas Vector Docs

A vector search MCP for document retrieval using MongoDB Atlas Vector Search and Voyage AI Context embeddings.

python

GitHub Stars

python

Language

4 months ago

First Indexed

3 weeks 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": {
    "patw-avs-docs-mcp": {
      "command": "uv",
      "args": [
        "run",
        "--with",
        "fastmcp, pymongo, requests",
        "fastmcp",
        "run",
        "<path to>/avs-docs-mcp/avs-mcp.py"
      ],
      "env": {
        "MONGODB_URI": "mongodb+srv://user:password@cluster.mongodb.net/?retryWrites=true&w=majority",
        "VOYAGE_API_KEY": "YOUR_VOYAGE_API_KEY",
        "VECTOR_DIMENSIONS": "256"
      }
    }
  }
}

You can run a FastMCP server that enables semantic document search over your Markdown documents. It ingests content, creates embeddings, stores data in MongoDB Atlas Vector Search, and exposes endpoints you can connect to with an MCP client to perform vector and lexical searches as well as retrieve full parent documents.

How to use

Your search workflow consists of ingesting documents, starting the MCP server, and querying with a client. Ingest content first to generate embeddings and organize documents with parent-child relationships. Then run the MCP server and connect your MCP client to perform semantic searches, lexical fallbacks, and full-document retrieval.

How to install

Prerequisites you need on your system before running the server:

Python 3.10 or newer
MongoDB Atlas cluster with vector search enabled
Voyage AI API key

Install steps you should follow in order:

Clone the project directory that contains the MCP server code and related scripts.
Install dependencies using Python’s package manager.
Create and configure a .env file with your credentials and settings.

Exact commands to run (copy and execute each line):

git clone https://github.com/patw/avs-docs-mcp.git
cd avs-docs-mcp
pip install -r requirements.txt
# Copy sample environment configuration to a real .env and fill in credentials
cp sample.env .env
# Edit .env to set MongoDB URI, database/collection names, Voyage AI API key, and vector dimensions

Configuration and tooling

Configure how the MCP server runs and how it is connected to clients. The server is designed to be started via an MCP runner, which in this example is invoked through a local runtime command. You will typically provide the runtime command, arguments, and any necessary environment variables.

Available tools

search_documents_vector

Primary search method using vector similarity; returns document chunks with metadata and similarity scores for semantic queries.

search_documents_lexicaly

Fallback search using lexical/text matching; returns full parent documents with search scores for exact or text-based queries.

get_parent_document

Retrieve the complete parent document by its ID to provide full context for a chunk.

Built by

VeilStrat

AI signals for GTM teams