Every integration.
One connection point.

The PLRX Agentic Execution Platform connects to the full infrastructure layer it needs — AI models, cloud services, observability, authentication, healthcare protocols, event streaming, and secrets management. Every one of those integrations is hidden from the AI client. From the outside, there is a single point of entry: the PLRX MCP Server.

PLRX MCP Server All Integrations

The universal
integration gateway.

The PLRX MCP Server is the single point through which any AI client — ChatGPT, Claude Code, Cursor, OpenCode, or any Model Context Protocol-compatible client — discovers and calls tools, reads resources, and receives streaming responses from external servers. The AI client needs zero knowledge of the underlying systems. It does not know which AI models are running. It does not know which protocols are in use. It does not know which infrastructure is behind the platform. It connects to the PLRX MCP Server once, receives a tool manifest, and invokes missions by name.

  • ▸
    Tool Discovery
    AI clients retrieve a structured tool manifest listing every available agent workflow — its name, parameters, and description. New agents appear in the manifest automatically on deployment. No client update required.
  • ▸
    Tool Invocation
    A single MCP tool call triggers a complete PLRX mission — initiating the orchestrator, delegating to specialist agents, coordinating with external systems, and tracking state through the PLRX Durable State Machine. The client receives a run ID immediately. The mission runs asynchronously to completion.
  • ▸
    Resource Reading
    Clients read structured resources — mission status, workflow state, agent health, execution logs — through the MCP resource protocol. No direct access to underlying databases or agent APIs required.
  • ▸
    Streaming Responses
    Real-time mission progress is streamed to the AI client as workflow state advances. The client receives state transition events — document received, eligibility confirmed, prior auth submitted — without polling. Streamable HTTP transport.
  • ▸
    Tenant-Scoped Visibility
    Tool visibility is scoped per tenant. A client authenticated for tenant A sees only tenant A's tools and resources. Cross-tenant leakage is architecturally impossible — not a configuration setting.
  • ▸
    OAuth2 Authentication
    Every MCP connection is authenticated via OAuth2 bearer tokens. Every tool call is logged to the WORM audit trail — caller identity, tool name, parameters, timestamp, and full response. No anonymous access.
MCP · tools/list · AI client discovers available missions
# AI client lists available tools on connect { "method": "tools/list" } ───────────────────────────────────────────────── { "tools": [ { "name": "mission_orchestration" "description": "Process a medical supply request end-to-end" "inputSchema": { "order_id": "string", "tenant_id": "string" } } { "name": "get_mission_status" "description": "Retrieve current state of a running mission" "inputSchema": { "run_id": "string" } } { "name": "process_rcm_claim" "description": "Submit and manage a revenue cycle claim" "inputSchema": { "claim_id": "string", "tenant_id": "string" } } ] }
MCP · tools/call · mission invoked by AI client
# Claude Code triggers a full PLRX mission { "method": "tools/call" "params": { "name": "mission_orchestration" "arguments": { "order_id": "ORD-787", "tenant_id": "tenant-abc" } } } ── Platform responds immediately ───────────────── { "content": [{ "type": "text", "text": "Mission #US-787 started" }] "run_id": "order-787-tenant-abc" } ── Streaming state updates follow ──────────────── MISSION_RECEIVED → document notification sent PHASE_VALIDATION → NPI valid · payer: ACTIVE AWAITING_INPUT → input_required · awaiting F2F
INTEGRATION ARCHITECTURE · AI CLIENTS → MCP SERVER → PLRX PLATFORM → UNDERLYING SYSTEMS
AI CLIENT
ChatGPT
AI CLIENT
Claude Code
AI CLIENT
Cursor
AI CLIENT
OpenCode
↓ ↓ ↓ ↓
MCP · Streamable HTTP · OAuth2
↓
A2A · JSON-RPC 2.0
↓ ↓ ↓ ↓
HTTPS · EDI · OpenTelemetry · Event streaming
AI MODELS
Claude Sonnet / Opus · Gemini Pro · ML
PROTOCOLS
FHIR R4 · EDI X12 · NPI
CLOUD
AWS · Kubernetes · Streaming
OBSERVABILITY
OpenTelemetry · Tracing · Logs

The full infrastructure layer.

Every component the PLRX Agentic Execution Platform connects to — organised by category. No proprietary buses, no black boxes. Each integration is a named protocol, a named provider, or an open standard.

Open standards only.

Every wire-level protocol used by the PLRX platform is an open standard. No proprietary formats. No vendor-specific message envelopes.

Protocol Category Purpose Used by

Every integration your autonomous operations needs.

Every integration listed here is live in production deployments today.

Request access to see the full integration layer running on your workflow.

Request Access Back to Platform →