groundworkers

groundworkers gives you read-only OMOP concept and mapping tools that can be used either over MCP or directly from Python.

Choose the interface that fits your app

Use the MCP server when you want tool discovery, process isolation, or a shared service for multiple clients.
Use the Python library when you want in-process calls with normal Python return values and exceptions.

At a glance

flowchart TD A[Python app] --> APP[build_application] B[MCP client] --> SERVER[create_server] APP --> S[services/] SERVER --> T[tools/] T --> S S --> AD[adapters/] AD --> OG[omop-graph] AD --> OE[omop-emb] AD --> DB[(OMOP DB)]

What most users care about

Concept and hierarchy lookup for deterministic OMOP operations
Search and grounding for lexical and semantic retrieval
Mapping workflows for candidate bundles, parent backoff, and context packets
One configuration model for both MCP and direct Python use

For direct Python consumers, the main entrypoint is build_application(config) and app.services.mapping.

Where to read next

Architecture for the package structure and layer boundaries
Integrations for end-to-end MCP and Python examples
Mapping Tools for mapping-oriented workflows
API Reference for MappingService and build_application

Relation to groundcrew

Info

groundcrew usually connects to groundworkers over MCP. If you are building a Python application, you can also import the same mapping logic directly through build_application(config).