Hecate is an MCP server tool designed to provide programmatic access to OMOP (Observational Medical Outcomes Partnership) vocabulary concepts. Developed by OHDSI/Hecate, this server acts as a semantic search engine for the OHDSI vocabulary. Built primarily with TypeScript, Hecate supports both local deployment as a JavaScript file and direct access via a hosted Streamable HTTP version. It integrates with various AI clients, enabling users to explore medical concepts and their relationships for diverse applications.

What does Hecate connect to?

Hecate MCP Server connects to the Hecate API, offering access to a vast collection of OMOP vocabulary concepts. This server is a component of the broader Hecate project, which functions as a specialized semantic search engine for the OHDSI vocabulary. Its core purpose is to bridge AI clients with structured medical knowledge, allowing for precise queries and data retrieval from established medical terminologies and classifications.

Core features and tools

The Hecate MCP Server exposes several callable tools for interacting with medical concepts. These include Search concepts for semantic queries across OMOP vocabularies like SNOMED, ICD-10, RxNorm, and LOINC. Users can also Lookup by ID to retrieve full metadata for specific OMOP concepts. Relationships allows exploration of cross-vocabulary mappings and standard concept connections, while PHOEBE relationships offers curated clinical relationships for phenotype and cohort definitions. Lastly, Hierarchy expansion enables navigation of concept hierarchies up and down.

Who is it for?

Hecate MCP Server serves users of AI clients, including Claude Desktop, Cursor, Windsurf, opencode, and ChatGPT (Plus, Team, or Pro), who require programmatic access to OMOP vocabulary concepts. It is suitable for those building applications or workflows that depend on precise medical terminology and concept relationships. Skip this if you do not require programmatic access to medical concepts through an AI client.

How to get started

No credentials are required for the Hecate MCP Server. The hosted version operates using Streamable HTTP, providing a direct communication channel for clients. For local deployments, Hecate functions as a stdio MCP server, establishing communication through standard input/output channels. This design allows for straightforward integration with compatible AI clients, including Claude Desktop, Cursor, Windsurf, opencode, and ChatGPT (Plus, Team, or Pro).

Use cases

• Performing semantic searches for medical concepts using natural language queries, allowing users to find relevant OMOP vocabulary entries across various domains like SNOMED, ICD-10, RxNorm, and LOINC.
• Retrieving full, detailed metadata for specific OMOP concepts by their unique ID, providing deep insights into their attributes and definitions.
• Exploring the intricate relationships and hierarchical structures between different medical concepts, which is essential for developing precise phenotype and cohort definitions in research and clinical settings.

Hecate offers a specialized gateway for AI clients to interact with OMOP vocabulary concepts, supporting semantic search and concept exploration. Its design supports both hosted and local deployments, integrating with popular AI platforms. The server provides essential tools for working with medical terminology and relationships. Start by performing a semantic search for medical concepts using a natural language query.