Agent Care MCP Server

Agent Care is a specialized Model Context Protocol (MCP) server that bridges conversational AI assistants with real‑world electronic medical records (EMRs).
Its core purpose is to give Claude Desktop and Goose Desktop users a secure, standardized way to query patient data from leading EMR platforms such as Cerner and Epic. By leveraging SMART on FHIR APIs, the server handles OAuth2 authentication, token exchange, and scope management automatically, so developers can focus on building clinical workflows instead of plumbing the intricacies of each vendor’s API.

The server exposes a rich set of FHIR‑centric tools that map directly to common clinical data entities: searching for patients, retrieving observations, conditions, medications, encounters, allergies, procedures, care team members, and care plans. In addition to basic lookups, it offers higher‑level functions like for arbitrary FHIR queries and for longitudinal lab data. These tools are designed to be null‑safe and cache responses, reducing latency and avoiding duplicate requests. On top of patient data, Agent Care integrates medical research services—PubMed article search, clinical trial discovery, and FDA drug‑interaction checks—providing a one‑stop shop for evidence‑based decision support.

For developers, the MCP server simplifies integration into AI workflows in several ways. First, it abstracts authentication: once a developer registers an app with Cerner or Epic and supplies the client credentials, the server manages token refreshes behind the scenes. Second, each tool is documented with clear parameter requirements (e.g., , , or ), enabling straightforward prompt construction. Third, the server’s response caching and error handling mean that conversational agents can reliably deliver consistent information without needing to implement custom retry logic. Finally, the server’s prompt templates and sample screenshots illustrate typical use cases—such as generating SOAP notes, creating encounter timelines, or conversing about a patient’s care plan—making it easier to prototype clinical assistants.

Key use cases include:

• Clinical decision support: A clinician asks an AI to summarize a patient’s recent vitals, medication list, and active conditions.
• Research assistance: A researcher queries PubMed for the latest trials on a rare disease and cross‑checks drug interactions.
• Workflow automation: An assistant drafts a SOAP note from structured FHIR data, then pushes it back to the EMR.
• Patient engagement: A virtual nurse pulls lab trends and explains results in plain language.

Agent Care’s standout features are its tight integration with both Cerner and Epic, comprehensive coverage of FHIR resources, built‑in research toolchain, and emphasis on secure, cache‑friendly interactions. For developers building AI assistants that need to talk to real patient data and evidence sources, this MCP server offers a ready‑made, well‑documented foundation that accelerates time to production while maintaining compliance with healthcare data standards.