Healthcare Context Intelligence Engine
v0.3.1A powerful MCP server built with NitroStack
Add via Cursor Settings UI (Settings > Features > MCP > Add New MCP Server):
{
"mcpServers": {
// your other mcp servers
"healthcare-context-intelligence-engine": {
"url": "https://synapsemed-a-healt-teamteam-amrita-university-amritapuri-campus.app.nitrocloud.ai/mcp"
}
}
}
Connect remote tools directly via Claude's Web UI:
Configure custom tools directly via ChatGPT's Web UI:
Add the following configuration block under mcpServers in your Antigravity configuration file (~/.gemini/config/mcp_config.json):
{
"mcpServers": {
// your other mcp servers
"healthcare-context-intelligence-engine": {
"serverUrl": "https://synapsemed-a-healt-teamteam-amrita-university-amritapuri-campus.app.nitrocloud.ai/mcp"
}
}
}
Add the following configuration block to your Codex configuration file (~/.codex/config.toml):
[mcp_servers.healthcare-context-intelligence-engine] url = "https://synapsemed-a-healt-teamteam-amrita-university-amritapuri-campus.app.nitrocloud.ai/mcp"
Connect directly using the Server-Sent Events endpoint:
https://synapsemed-a-healt-teamteam-amrita-university-amritapuri-campus.app.nitrocloud.ai/mcp
Classify emergency type and severity hint from raw text input (non-diagnostic)
Parse raw input into structured emergency orchestration intent
Extract prioritized actionable goals from classified intent
Build a complete emergency execution plan from raw input. Runs discovery, retrieval, context, generation, and validation concurrently. Returns a plan object requiring human approval before execution-class actions.
Retrieve normalized patient summary for emergency orchestration (non-diagnostic)
List patient allergies by patient_id
List active medications by patient_id
Search and rank hospitals by query, location, and filters
Find nearest hospital matching required departments
List departments available at a hospital
Resolve location text to lat/long coordinates
Calculate distance in km between origin and destination
Calculate route and ETA between origin and destination
Compute weighted context intelligence score for a retrieval item
Hybrid retrieve and score hospital candidates
Re-rank retrieval results by rerank_score descending
Build context graph from intent, discovery, and ranked hospitals
Infer priority resource flags from intent and discovery context
Build an LLM-ready prompt from intent and compressed context. Includes non-diagnostic constraints. Returns the prompt string and system prompt.
Compress a full HCIE context (patient, hospitals, contacts, maps) into a token-efficient summary. Preserves: allergies, emergency contacts, symptoms (always). Boosts: location + hospital for dispatch intent.
Validate an ExecutionPlan for structural correctness: required fields, non-empty steps, valid action names, correct step ordering.
Hard safety gate — scans plan text for diagnostic or prescriptive language. Returns safe=false and the offending terms if any diagnostic language is detected. This gate is enforced at code level and cannot be bypassed.
Generate a concise, user-facing, non-diagnostic explanation of an ExecutionPlan. Summarizes steps in plain English without clinical language.
Human approval gate for an ExecutionPlan. Call with approved=true to unlock execution-class actions (call_ambulance, connect_physician). Call with approved=false to deny the plan. This gate is enforced at code level — execution-class tools will throw if called without approval.
🚑 EXECUTION-CLASS — Dispatch an ambulance to the patient location. Requires confirm_plan(plan_id, approved=true) to have been called first. Will throw an error if the plan is not approved.
👨⚕️ EXECUTION-CLASS — Connect to an available physician for the case. Requires confirm_plan(plan_id, approved=true) to have been called first. Will throw an error if the plan is not approved.
Send emergency notification to a family/emergency contact by contact_id.
Generate navigation instructions from current location to a hospital.
Share the case location with a recipient (hospital, contact, dispatcher).
Share a non-diagnostic patient context summary with the receiving hospital or physician.