Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@Prior Authorization System MCP Server check status of PA request #12345"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
End-to-end Prior Authorization decision-support system with a multi-agent AI pipeline, MCP Tool Registry, and PostgreSQL backend.
Original wireframe: https://www.figma.com/design/5qn3BZ9VfJ2eoa06sytSY0/Prior-Authorization-Wireframe
Frontend: React 18 + Vite + Tailwind CSS + shadcn/ui
Backend: Express.js (ESM) with JWT authentication
Database: PostgreSQL 16 (6 tables, 400+ seed rows)
AI Pipeline: 4 agents + 6 MCP tools + 9-step processing pipeline
Related MCP server: CzechMedMCP
Node.js 18+
PostgreSQL 16+
npm
# 1. Install frontend dependencies
npm install
# 2. Install backend dependencies
cd server && npm install && cd ..
# 3. Create database
createdb prior_auth_db
# 4. Run schema + seed data (15 members, 75 claims, 15 policies, 15 PA requests)
node server/db/setup.js
# 5. Create .env.local for frontend API URL
echo "VITE_API_BASE_URL=http://localhost:3001/api/v1" > .env.local# Start backend (port 3001)
node server/index.js
# Start frontend (port 5173) — in a new terminal
npm run devUsername | Password | Role |
admin | admin123 | Medical Director |
nurse | nurse123 | Clinical Reviewer |
analyst | analyst123 | Claims Analyst |
# Health check
curl http://localhost:3001/api/v1/health
# List MCP tools
curl http://localhost:3001/api/v1/mcp/tools
# List PA requests
curl http://localhost:3001/api/v1/requestsnode server/db/setup.js --resetSee DESIGN.md for full architecture documentation covering database schema, MCP tool registry, agent pipeline, API routes, and frontend integration.
This server cannot be deployed
Human-in-the-loop for AI agents over MCP: durable approvals with a hosted review page & audit trail
Stateless agentic tools over MCP: concept extraction, long-context, knowledge graph, planning.
MCP-Native LLM Orchestration Agent
MCP Hub: AI service discovery, per-user OAuth, and multi-service workflow orchestration
