# ferrants/memvid-mcp-server [Health: Active]

**Category:** 🗄️ Databases  
**Repository:** https://github.com/ferrants/memvid-mcp-server  
**GitHub Stars:** 10  
**Views:** 3  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/ferrants-memvid-mcp-server

## Description
Python Streamable HTTP Server you can run locally to interact with memvid storage and semantic search.

## Claude Desktop Quick Installation
Heuristic fallback — verify the package name and runner against the repository README before running it. Uses `uvx` (confidence: low):

```json
"mcpServers": {
  "memvid-mcp-server": {
    "command": "uvx",
    "args": ["ferrants-memvid-mcp-server"]
  }
}
```

## Documentation

## What ferrants/memvid-mcp-server MCP server does

The ferrants/memvid-mcp-server MCP server connects MCP clients to memvid, a storage approach that encodes text data into videos for later lookup. It runs as a local Streamable HTTP server and makes two actions available: adding text chunks and searching for relevant chunks.

The server is suited to small, local text collections that need semantic retrieval through an MCP-compatible client. It does not describe general-purpose database operations, document management, or remote hosting. Its documented scope is limited to writing chunks into memvid storage and querying that stored content.

## How it works

When chunks are added, the server uses memvid to encode the text into a memory video. A search request then queries the stored content and returns the top-matching chunks. Search returns five results by default, and callers can change that count with the `top_k` parameter.

The repository describes the server as Streamable HTTP rather than a standard local stdio process. After startup, an MCP client connects to the HTTP address exposed by the Python process. The example configuration uses `http://localhost:3000` and identifies the transport as `streamable-http`.

## Setup and configuration

The ferrants/memvid-mcp-server MCP server requires Python 3.11 in the documented setup. Create and activate a virtual environment, install the dependencies from `requirements.txt`, and start the process with `python server.py`:

```bash
python3.11 -m venv my_env
. ./my_env/bin/activate
pip install -r requirements.txt
python server.py
```

The default connection example points clients at port 3000. To use another port, set the `PORT` environment variable when starting the process, such as `PORT=3002 python server.py`. Configure the MCP client with the server URL and the `streamable-http` transport type.

## Tools and capabilities

- Add text chunks to memvid-backed storage with `add_chunks`.
- Search stored content with `search`.
- Retrieve the top five matching chunks by default.
- Set a different result count through the `top_k` search parameter.
- Connect over a local Streamable HTTP endpoint.

## Limitations and notes

Adding chunks resets `memory.mp4` each time, according to the repository documentation. The project notes that incremental additions may not currently be supported or may not have a known implementation. Existing memory-video contents should therefore be treated carefully when adding new chunks.

The provided material documents local execution and a localhost client configuration only. It does not specify authentication, authorization, persistence beyond the generated memory video, deployment instructions, or named client integrations. The server also depends on the repository's Python requirements, which must be installed before startup.

_Full upstream README: https://allmcps.com/mcp/ferrants-memvid-mcp-server/readme_