MCP Server

The iTechSmart UAIO MCP server at https://mcp.itechsmart.dev exposes the autonomous operations platform to AI agents over the Model Context Protocol — 66 tools, Bearer authentication, per-key scopes, rate limiting, and a ProofLink receipt sealed for governed actions. It also speaks the Linux Foundation A2A (Agent-to-Agent) protocol on the same surface.

Live server status (from GET /health, 2026-07-01):

{
  "status": "ok",
  "service": "itechsmart-uaio-mcp",
  "version": "2.2.0",
  "transport": "http+sse",
  "tools": 66,
  "auth_required": true,
  "rate_limit_per_min": 60,
  "rate_limit_simulate_per_min": 10
}

Endpoints

Method	Path	Auth	Purpose
GET	`/health`	No	Server status, tool count, session count
GET	`/mcp/tools`	No	Machine-readable tool registry (JSON)
GET	`/sse`	Yes	MCP SSE transport — connect here
POST	`/messages`	Yes	MCP message channel for SSE sessions
GET	`/.well-known/agent.json`	No	A2A agent card
POST	`/a2a`	Yes	A2A JSON-RPC (message/send)

Authentication is a Bearer API key (or ?api_key= query parameter), validated at connect time and fail-closed. Keys carry scopes — full keys reach all 66 tools; read-only audit keys are limited to the prooflink.*, mission.* and compliance.* tools. Rate limits: 60 req/min (10/min for simulate_infrastructure_attack). To get a key: [email protected]. Details on the Authentication page.

Connect: Claude Code

claude mcp add --transport sse itechsmart https://mcp.itechsmart.dev/sse --header "Authorization: Bearer YOUR_MCP_KEY"

Then run claude and the itechsmart tools appear automatically (e.g. query_uaio_status, verify_prooflink_receipt). Remove with claude mcp remove itechsmart.

Connect: Cursor

Add to .cursor/mcp.json (project) or ~/.cursor/mcp.json (global):

{
  "mcpServers": {
    "itechsmart": {
      "url": "https://mcp.itechsmart.dev/sse",
      "headers": { "Authorization": "Bearer YOUR_MCP_KEY" }
    }
  }
}

Connect: any SSE / HTTP client

The server implements the standard MCP SSE transport. Open the event stream, read the endpoint event to learn your session’s message URL, then POST JSON-RPC MCP messages to it:

# 1) open the SSE stream (keep it open)
curl -N -H "Authorization: Bearer YOUR_MCP_KEY" https://mcp.itechsmart.dev/sse
# → event: endpoint
#   data: /messages?sessionId=<id>

# 2) in another shell, list tools over the session channel
curl -X POST "https://mcp.itechsmart.dev/messages?sessionId=<id>" \
  -H "Authorization: Bearer YOUR_MCP_KEY" -H "Content-Type: application/json" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'

# 3) call a tool
curl -X POST "https://mcp.itechsmart.dev/messages?sessionId=<id>" \
  -H "Authorization: Bearer YOUR_MCP_KEY" -H "Content-Type: application/json" \
  -d '{"jsonrpc":"2.0","id":2,"method":"tools/call","params":{"name":"query_uaio_status","arguments":{}}}'
# tool results arrive as events on the SSE stream from step 1

The full tool registry with input schemas is machine-readable at mcp.itechsmart.dev/mcp/tools (no auth).

A2A protocol (agent-to-agent)

The same governed surface is exposed over A2A. The public agent card describes the interface:

curl https://mcp.itechsmart.dev/.well-known/agent.json

{
  "protocolVersion": "0.3.0",
  "name": "iTechSmart UAIO Agent",
  "url": "https://mcp.itechsmart.dev/a2a",
  "preferredTransport": "JSONRPC",
  "version": "2.2.0",
  "provider": {"organization": "iTechSmart", "url": "https://itechsmart.dev"},
  "documentationUrl": "https://mcp.itechsmart.dev/mcp/tools",
  "securitySchemes": {"bearer": {"type": "http", "scheme": "bearer",
    "description": "iTechSmart MCP API key as Bearer token"}}
}

Send messages with JSON-RPC message/send (same Bearer key). Plain-text parts route to the OctoAI reasoning pipeline; a structured DataPart invokes a named skill (the same skills as the MCP tools):

curl -X POST https://mcp.itechsmart.dev/a2a \
  -H "Authorization: Bearer YOUR_MCP_KEY" -H "Content-Type: application/json" \
  -d '{
    "jsonrpc": "2.0", "id": 1, "method": "message/send",
    "params": {"message": {"role": "user", "kind": "message", "messageId": "m1",
      "parts": [{"kind": "text", "text": "What is the current UAIO platform status?"}]}}
  }'

Tool reference — all 66 tools

Grouped by capability. Names and descriptions come from the server’s live tool registry (v2.2.0); input schemas for each are at /mcp/tools.

ProofLink & receipt verification (5)

Tool	Description
`verify_prooflink_receipt`	Verify the cryptographic integrity of a ProofLink receipt from iTechSmart's autonomous IT operations platform.
`get_receipt_chain`	Fetch and verify the complete ProofLink receipt chain from iTechSmart's production ledger.
`prooflink.verify_chain`	Read-only audit scope: verify the ProofLink receipt chain for EU AI Act Article 12, CISO, and auditor workflows.
`prooflink.verify_receipt`	Read-only audit scope: verify one ProofLink receipt by ID without granting mutation rights.
`prooflink.search_receipts`	Read-only audit scope: search recent ProofLink receipts for ledger-backed post-hoc reconstruction.

Read-only audit & mission scopes (3)

Tool	Description
`mission.list_incidents`	Read-only mission scope: list recent autonomous IT incidents for audit and operations review.
`mission.cluster_health`	Read-only mission scope: return live cluster health without granting execution or remediation rights.
`compliance.audit_summary`	Read-only compliance scope: return the live compliance audit summary for CISO/auditor review.

Platform status & incidents (8)

Tool	Description
`query_uaio_status`	Get the current operational status of the iTechSmart UAIO (Unified Autonomous IT Operations) platform.
`get_platform_briefing`	Get a full real-time platform health briefing for the iTechSmart UAIO platform.
`list_recent_incidents`	List recent autonomous IT remediation incidents from the iTechSmart UAIO platform.
`get_incident_details`	Fetch details of a specific autonomous remediation incident including the ProofLink receipt, remediation actions taken, before/after system state, time to detect, time to remediate, and NIST contro...
`simulate_infrastructure_attack`	Trigger a simulated infrastructure attack on the iTechSmart break-it sandbox to demonstrate the UAIO autonomous loop.
`dispatch_ag2_incident`	Dispatch an IT incident to the iTechSmart AG2 6-agent GroupChat for autonomous diagnosis.
`invoke_octoai_pipeline`	Invoke the OctoAI 7-node cognitive pipeline with a free-form prompt.
`search_platform_logs`	Search live systemd journal logs for any iTechSmart service.

Security & self-improvement (SIE / iSELF) (4)

Tool	Description
`get_sie_queue`	Retrieve the current SIE (Self-Improving Engine) security finding queue.
`approve_sie_finding`	Approve a specific SIE (Self-Improving Engine) security finding by queue index.
`trigger_sie_scan`	Kick off a fresh SIE (Self-Improving Engine) scan of the iTechSmart platform.
`get_iself_journal`	Retrieve the iSELF (Self-Healing Loop Framework) healing history.

Compliance (5)

Tool	Description
`get_compliance_status`	Get live compliance scores for the iTechSmart platform.
`probo_health`	Check Probo compliance platform health.
`probo_controls`	List all Probo compliance controls with status, category, and description (SOC2/EU AI Act).
`probo_risks`	List all Probo compliance risks with severity level and status.
`probo_summary`	Get Probo compliance aggregate summary: control pass/fail, risk high/medium/low, and task open/done counts.

Integrations & observability (7)

Tool	Description
`integration_status`	Parallel health check across all wired integrations: Langfuse, RAGflow, Shuffle, Tactical RMM, MeshCentral, and Probo compliance.
`langfuse_health`	Check Langfuse LLM observability platform health.
`langfuse_trace`	Create a Langfuse trace to log an LLM interaction (input, output, model, metadata).
`ragflow_health`	Check RAGflow RAG platform health.
`ragflow_query`	Query the RAGflow knowledge base with a natural-language question.
`shuffle_health`	Check Shuffle SOAR platform health.
`shuffle_trigger`	Trigger a Shuffle SOAR workflow by workflow ID with an optional request body.

Device & endpoint management (Tactical RMM / MeshCentral) (6)

Tool	Description
`trmm_health`	Check Tactical RMM health.
`trmm_agents`	List all Tactical RMM managed agents with status, OS, site, and last-seen info.
`trmm_summary`	Get Tactical RMM agent summary: total, online, offline, overdue counts, and site list.
`trmm_run_script`	Run a saved script on a Tactical RMM agent by agent ID and script ID.
`mesh_health`	Check MeshCentral remote device management health.
`mesh_devices`	List all devices enrolled in MeshCentral with hostname, OS, connectivity, and group info.

Learning Arbiter (3)

Tool	Description
`get_learning_queue`	Get patterns pending human approval from Learning Arbiter (Tier 2 items, score 50-85).
`approve_learning_item`	Approve a Tier 2 learning pattern for promotion to Agent Brain.
`get_learning_metrics`	Get Learning Arbiter cycle metrics: processed, promoted, rejected, known patterns.

CRM (GoHighLevel) (6)

Tool	Description
`ghl_create_contact`	Create a contact in GoHighLevel CRM.
`ghl_update_contact`	Update GHL contact fields.
`ghl_send_email`	Send email to GHL contact via Judge gate (held if flagged).
`ghl_trigger_workflow`	Trigger a GHL workflow for a contact.
`ghl_book_appointment`	Book a calendar appointment for a GHL contact.
`ghl_move_pipeline`	Move a GHL opportunity to a new pipeline stage.

Voice & audio (Deepgram) (3)

Tool	Description
`deepgram_tts`	Convert text to speech via Deepgram Aura (aura-asteria-en).
`deepgram_transcribe`	Transcribe audio from URL or file path using Deepgram Nova-2.
`deepgram_blog_audio`	Generate MP3 audio version of a blog post via Deepgram TTS.

ITSM (16)

Tool	Description
`itsm_create_ticket`	Create an ITSM ticket for an incident.
`itsm_get_ticket`	Retrieve a single ITSM ticket by ID, including embedded ProofLink receipt if sealed.
`itsm_list_tickets`	List ITSM tickets.
`itsm_close_ticket`	Close and resolve an ITSM ticket with a resolution note.
`itsm_run_loop`	Trigger the full 6-step ITSM ProofLink loop: DETECT -> TICKET -> DIAGNOSE -> REMEDIATE -> VERIFY -> CLOSE+SEAL.
`itsm_run_test`	Inject a synthetic test incident through the full ITSM ProofLink loop.
`itsm_get_mttr_stats`	Get Mean Time To Remediate (MTTR) statistics from resolved ITSM tickets.
`itsm_embed_receipt`	Embed a ProofLink SHA-256 receipt directly inside an existing ITSM ticket body.
`itsm_verify_receipt`	Verify the ProofLink receipt embedded in an ITSM ticket.
`itsm_list_open_incidents`	List all currently open incidents (ITSM tickets with status=open), sorted by severity.
`itsm_get_incident_timeline`	Get the full lifecycle timeline of an incident: detected, ticketed, diagnosed, remediated, verified, sealed.
`itsm_trigger_detection`	Run all 5 incident detectors (iSELF, Prometheus, Wazuh, NetworkScanner, OctoAI) and return raw findings without creating tickets.
`itsm_get_detection_sources`	Return status and last-scan results for all 5 ITSM incident detection sources.
`itsm_get_remediation_history`	Get the history of all auto-remediation actions, with MTTR and success rate.
`itsm_register_webhook`	Register a webhook URL to receive real-time ITSM ticket state changes (OPENED, UPDATED, RESOLVED).
`itsm_dashboard_summary`	Get the ITSM dashboard summary: open count, resolved today, avg MTTR, top failing services, and ProofLink sealing rate.

Governance and receipts

Mutating tools are mediated by the Arbiter constitutional governance engine, and governed actions seal SHA-256 ProofLink receipts to the public ledger — verify any of them at verify.itechsmart.dev or programmatically via the ProofLink API. Audit-scope keys let a CISO or external auditor use the prooflink.* / mission.* / compliance.* tools without any mutation rights.