Overview
What is MCP KIPRIS?
MCP KIPRIS is an MCP server that provides patent search tools using the Korean Intellectual Property Office (KIPRIS) API. It enables retrieval of both Korean and international patent information for AI assistants like Claude.
How to use MCP KIPRIS?
Set the KIPRIS_API_KEY environment variable, install dependencies (pip install -r requirements.txt), then run the server in HTTP mode (python -m mcp_kipris.sse_server --http --port 6274 --host 0.0.0.0) or stdio mode (python -m mcp_kipris.server). Use tool calls like patent_applicant_search with arguments such as applicant name and sort options.
Key features of MCP KIPRIS
- Korean patent search by applicant, keyword, application number, and righter.
- Detailed patent info and summary retrieval by application number.
- International patent search by applicant, keyword, application number, etc.
- Supports multiple country codes (US, EP, JP, CN, etc.).
- Sort options: publication date, application date, registration date, etc.
- All responses are returned in JSON format.
Use cases of MCP KIPRIS
- Search for recent patents by a specific applicant (e.g., Samsung Electronics).
- Find patents by keyword with sorting and filtering.
- Retrieve detailed or summary information for a given patent application number.
- Search international patents by applicant, keyword, or application number across multiple countries.
FAQ from MCP KIPRIS
What runtime dependencies does MCP KIPRIS require?
Python 3.12 or higher is required. Install dependencies with pip install -r requirements.txt or pip install -e ..
How do I obtain and set the API key?
Export the KIPRIS API key as an environment variable: export KIPRIS_API_KEY="your_api_key".
What transport modes are available?
The server supports both HTTP (SSE) mode and stdio mode, configured by command-line arguments.
What patent status codes are supported?
Status codes include: A (published), C (corrected publication), F (laid-open), G (corrected laid-open), I (invalid notice), J (cancellation notice), R (re-publication).
Are there any known limits or restrictions?
The README does not mention specific rate limits or data retention policies. The server provides DEBUG-level logging for monitoring tool calls.
