Skip to main content
IPRATE

Free public MCP

European IP data for AI assistants

Connect without an account or API key. The server selects bounded answers from one completed set of IPRATE static release assets, and states the exact release, source files, coverage, and limitations in every result.

Endpoint

https://mcp.iprate.eu/mcp

Transport: Streamable HTTP · Authentication: none · Operations: read-only

Set up your assistant

Claude Code

Run this command in your terminal. Open Claude Code and use /mcp to check the connection.

claude mcp add --transport http iprate https://mcp.iprate.eu/mcp

Documentation checked 11 October 2026. Claude Code instructions.

Claude Desktop and claude.ai

Open Customize, then Connectors. Choose Add custom connector. Name it IPRATE and paste the endpoint below. Choose No sign in. Enable the connector in your conversation.

https://mcp.iprate.eu/mcp

Team and Enterprise owners add the connector in Organization settings, under Connectors.

Documentation checked 11 October 2026. Claude connector instructions.

Cursor

Add this entry to .cursor/mcp.json in your project. Keep any existing server entries. Restart Cursor and check the MCP tools list.

{
  "mcpServers": {
    "iprate": {
      "url": "https://mcp.iprate.eu/mcp"
    }
  }
}

Documentation checked 11 October 2026. Cursor instructions.

ChatGPT

Open Plugins. Choose the plus button, then Add custom MCP server. Enter IPRATE as the name and paste the endpoint. Select anonymous access and create the plugin. Review the tools, then select IPRATE with @ in a new conversation.

https://mcp.iprate.eu/mcp

Workspace policy controls access. Ask your Business or Enterprise administrator to make the connection available to your workspace. The search and fetch tools support research over published IPRATE records.

Documentation checked 11 October 2026. ChatGPT connection instructions.

OpenAI Responses API

Install the OpenAI Python SDK. Set OPENAI_API_KEY and set OPENAI_MODEL to a model in your account that supports remote MCP. IPRATE access is free. OpenAI API usage is billed by OpenAI.

import os
from openai import OpenAI

client = OpenAI()
response = client.responses.create(
    model=os.environ["OPENAI_MODEL"],
    input="Which trademark firms have published ratings in Lithuania?",
    tools=[{
        "type": "mcp",
        "server_label": "iprate",
        "server_url": "https://mcp.iprate.eu/mcp",
        "require_approval": "never",
    }],
)
print(response.output_text)

The example allows the public read-only tools to run without an approval prompt.

Documentation checked 11 October 2026. OpenAI MCP instructions.

Six tools

find_ip_representatives

Find up to five released firms or attorneys by name, market, right type, leading Nice class, or leading-client evidence.

get_ip_representative_profile

Retrieve one released representative's bounded evidence profile and cohort results.

get_ip_market_snapshot

Return one already-computed market snapshot and up to five leading released firms.

get_iprate_coverage

Check which jurisdiction, right type, route, and window combinations exist in the selected release.

search

Return up to ten citable documents (profiles, market snapshots, coverage summaries) for an open research question.

fetch

Read one document returned by search in full, with its public URL and release metadata.

Good first prompts

  • “Which firms lead the released national trademark ranking for Lithuania?”
  • “Find released trademark representative profiles with Nice class 9 among their leading classes.”
  • “Which patent-market snapshots are present for Germany in the current release?”

What the free server covers

Each answer lists up to five representatives and five Nice classes. A search returns up to ten documents. Each result comes from a published release. You can choose the long or the recent window. The server only reads data. It never changes a record, places an order or takes payment. For register search, portfolio watches, bulk export and paging, use the Data API. The server gives no legal opinion.

Register names and returned text are quoted source records. Read them as data. They carry no instructions for your assistant. A filing record shows recorded activity. For current availability and legal status, check the official register.

Delivery samples

Weekly digest · Digest CSV

Need a larger extract, portfolio analysis, monitoring feed, or a custom question? Request analysis. Read the methodology, terms, and privacy policy.

The adapter source, schemas, and tests are Apache-2.0. The adapter contains no database driver, SQL, database credential, or application-API fallback. It reads released static JSON only; IPRATE's database and internal services are not included.