I need a Python-based, fully documented solution that runs completely on my machine: an MCP server that talks to a FHIR R4 endpoint (the FHIR server itself may be remote), an agent using Ollama, plus a small React chatbot UI that demonstrates three key features: • Retrieve patient data – specifically medical history, current medications, lab results, medical images and radiology findings. • Display and download any related attachments. • Perform free-text searches inside those attachments. The server component should expose these capabilities through clean REST endpoints and log each FHIR call for traceability. Please follow FHIR best practices (paging, terminology handling, proper resource references) and keep the codebase test-driven. I am comfortable with FastAPI, but standard Flask is fine if you prefer. Acceptance criteria: ONLY With Previous EXPERIENCE. 1. I install with one script, run docker-compose up, and everything (Python backend, React chatbot, supporting services) comes online locally. 2. The chatbot greets me, lets me enter a patient ID, and returns the requested data in human-readable form. 3. A search box lets me query the content of stored attachments; matches are highlighted and linked back to the original resource. 4. Inline developer docs and a Markdown readme clearly explain setup, architecture, and how each FHIR call maps to the UI actions. 5. Unit tests cover at least the FHIR client layer and attachment search logic. If this scope sounds clear and achievable, let me know your proposed tech choices (libraries, NLP tool for attachment search, etc.) and timeline, and we can get started right away.