OHS labs

Repository for defining quick AI experiments that could be rolled into future OHS features or capabilities

Current explorations include:

1. Natural language querying of the OHS Data Pipes DWH using AI

• Build skill (or set of skills) for retrieving context from the FHIR Implementation Guide or configs, building the query and enforcing guardrails (e.g. can't DELETE) etc
• Leverage MCP for database layer execution
• Could package this up into a workflow

2. Agentic workflow for deploying OHS FHIR Data Pipes

• Develop skill for interacting with FHIR Data Pipes
• Can leverage existing APIs and documentation
• Demonstrate an agentic workflow

3. Skills for working with kotlin-fhir models

• Develop CLI wrapper to interact with the kotlin-fhir models
• Define skills from the API documentation to ensure validation etc
• Demonstrate agentic use case of a chat app working with FHIR data via the CLI