Patent Connector MCP Server - New Release v1.1.0

My consulting philosophy: never talk about something you don't fully understand and know. So when MCP came up as a topic, I built a patent MCP server to get real implementation experience. I also found the concept behind MCP very interesting - a standardized protocol for giving LLMs access to external data, which works well for patent information where data is spread across multiple offices and APIs.

Since then the server has been picking up users steadily - people sign up daily without any advertising. So I thought it was worth investing in a proper update.

If you're already connected, reload the connector in your chatbot settings to pick up all the new tools.

What's New

DPMA Provider - German Patent Office

The server now connects to the German Patent and Trademark Office (DPMA) via the DPMAconnect Plus API. German patent numbers (DE...) are automatically routed to DPMA, with EPO as fallback. This brings the total to three patent office providers: EPO, USPTO, and DPMA.

Compound Tools

Instead of having the LLM chain multiple tool calls to assemble information, this release adds tools that combine several lookups in one call:

• get_patent_summary - bibliography, legal status, family size and countries, patent term/expiry, and current owner - all in one request, running concurrently
• get_prosecution_timeline - chronological prosecution history with key milestones (filed, published, examination requested, granted) combined from bibliography, register events, and legal status
• compare_patents - side-by-side comparison of 2-10 patents across any provider

These return partial results when some lookups fail, so you always get what's available.

More Tools From the Client Libraries

This release exposes more of the underlying open source client libraries I maintain:

• get_continuity - parent/child application chains (USPTO)
• get_assignments - ownership transfer history (USPTO)
• get_register_events - prosecution events with categorization (EPO/USPTO)
• get_patent_term - term calculation and expiry dates
• convert_patent_number - convert between original, epodoc, and docdb formats (EPO)
• get_classification - CPC/IPC classification hierarchy lookup
• get_search_syntax - per-provider query documentation
• PTAB tools - search proceedings, get proceeding details, search decisions

That brings the total to 22 tools across the three providers.

Self-Hosted Deployment

The whole stack can now run on-premises. Law firms, corporate IP departments, and R&D teams can host their own LLM (using AnythingLLM or similar), connect this self-hosted MCP server, and have the entire processing pipeline between their local network and the patent offices. No third party involved.

Authentication Improvements

The authentication system now supports dual auth mode - both direct API access and browser-based OAuth flows. This not only makes self-hosted deployments work properly but is also a prerequisite for submitting the server as an official ChatGPT app, so it will be available as an official ChatGPT app soon.

What's Next?

This is still somewhat experimental, so if you run into bugs or have feature ideas, reach out any time.

A few things I'm curious about:

• Should JPO (Japan Patent Office) be the next client library and MCP integration?
• Are there other compound tools that would save you time - something you find yourself asking the LLM to piece together repeatedly?
• What workflows would benefit most from structured patent data access?

The client libraries are open source.

The MCP server is live at patent.dev/patent-connector