CIViC MCP: Model Context Protocol Server and Evaluation Data for Integrating Large Language Models with the CIViC Knowledgebase

Summary: The Clinical Interpretation of Variants in Cancer (CIViC) knowledgebase provides a community-driven, open-source platform for discussing the biological and Clinical Significance of molecular variants in cancer. To enable users to make complex connections between CIViC information, we developed the CIViC Model Context Protocol (MCP) server, which allows large language models (LLMs) to directly interface with the CIViC API through natural language, facilitating the rapid summarization of expertly curated cancer variant interpretations. This repository includes the open-source CIViC MCP server and the evaluation data used for benchmarking its performance (including the eval_QA_experiment dataset). Together, these resources support reproducible experimentation with LLMs for biomedical information retrieval and evidence summarization. Contents: MCP server implementation (/civic-mcp-server) Evaluation scripts and results (/eval_QA_experiment) Example GraphQL queries and response templates Documentation for server setup and usage