MedScribe MCP

MedScribe uses MCP to bridge PromptOpinion with FHIR. Doctors use text commands to pull live patient lists and scribe notes directly in chat, eliminating the need to create or upload external PDFs.

Comment

Inspiration

Speeding up the clinical workflow. While test reports and discharge summaries belong in PDFs, daily progress notes shouldn't be trapped in slow, manual uploads. We wanted a "fast lane" for immediate clinical scribing.

What it does

MedScribe is a specialized MCP bridge for instant clinical documentation. It handles the "active" part of a doctor's day so they don't have to generate a PDF for every quick update. The Problem: Creating and uploading a PDF for a simple 2-line patient update is a waste of clinical time. The Solution: A direct-to-data text bridge. Doctors view patient lists and save progress notes instantly via chat, leaving the heavy PDF uploads for long-form reports.

How we built it

Protocol: Model Context Protocol (MCP) for the real-time bridge. Backend: .NET 10 Data: Structured JSON storage designed for easy FHIR mapping. Platform : PromptOpinion AI

Challenges we ran into

Separating Concerns. The hardest part was ensuring the AI understood when to query the live record vs. when to refer to an uploaded document. We tuned the tool definitions so the AI knows exactly when to use the "Fast Lane" (MedScribe).

Accomplishments that we're proud of

Zero-Friction Updates. We successfully created a workflow where a doctor can update a patient's status in seconds. No saving, no naming files, no uploading—just pure communication.

What we learned

We learned that AI is most powerful when it has direct access. By giving the AI a "live" connection to the patient list, it becomes a partner that understands context rather than just a document reader.

What's next for MedScribe MCP

Embedded MCP Interface: Building a custom UI inside the agent window to provide a "single-pane-of-glass" experience where doctors can see patient data and the chat simultaneously.

Intelligent Clinical Analysis: Adding a layer of analytics to scan saved notes for trends, potential medication risks, or gaps in care, turning raw text into proactive clinical insights.

Unified File Intake: Streamlining the submission of PDFs and lab reports directly through the chat, allowing the AI to instantly parse and attach external documents to the structured patient record.

Built With

Share this project:

Updates