PatentSafe MCP

by amphora

213 downloads
Not rated
GitHub

About

This is a model context protocol server that connects to PatentSafe and retrieves documents.

Details

Author
amphora
Downloads
213
Categories
Other

- Connects to PatentSafe to retrieve documents.
- Uses raw Lucene queries against the PatentSafe database.
- Includes a --max-chars flag to limit character output (default 500k).
- Works with any MCP client.
- Installed via uvx from a GitHub repository.

Configure the server in your MCP client (tested with Claude) by specifying the uvx command with the server package from GitHub, the PatentSafe base URL, and your authentication token. The server runs raw Lucene queries against the PatentSafe database.

PatentSafe MCP

This is a MCP server for PatentSafe. It is used to connect to PatentSafe and retrieve documents.

Installation

For Claude, you can install the MCP server having a ~/Users/simonc~/Library/Application Support/Claude/claude_desktop_config.jsonfile which looks something like this on macOS with Homebrew and uv installed:

{
  "mcpServers": {
    "USEFULNAME": {
      "command": "/opt/homebrew/bin/uvx",
      "args": [
        "--from",
        "git+ssh://git@github.com/amphora/patentsafe-mcp.git",
        "patentsafe-mcp",
        "PATENTSAFE_BASE_URL",
        "PATENTSAFE_AUTH_TOKEN"
      ]
    }
  }
}

You'll need to replace:

- USEFULNAME with a useful name for the server (e.g. patentsafe)
- PATENTSAFE_BASE_URL with the base URL of your PatentSafe instance (e.g. https://demo.morescience.com)
- PATENTSAFE_AUTH_TOKEN with your PatentSafe authentication token you can get from your user settings.

Usage

We've tested it with Claude, but it should work with any MCP client.

How it works (and debugging)

The MCP server does raw Lucene queries against the PatentSafe database, which is not something users often do. If you want to do a Lucene query, you can use the following URL:

https://PATENTSAFE_BASE_URL/ps/read/search.html?pq=1

Limits

Our initial testing shows that PatentSafe can easily return more documents than the LLM can process. So there's a --max-chars flag to limit the number of characters returned, default 500k.

Support

This is very much a work in progress. It works for us, and will likely work for you, but it's not production ready and is under active development.

You can raise issues directly with Amphora's support team in the usual way, or contribute to this GitHub project with pull requests, issues, discussion etc.

Future Work

- Add support for exposing Metadata fields
- Move the MCO server natively into PatentSafe (if that's the way clients evolve)

No reviews yet — be the first

Sign in to leave a review

Use Google, GitHub, or an email account so ratings stay tied to real people.

Email sign in

No reviews posted yet.