Stexa Voice Mcp

What is Stexa Voice MCP?

A voice telephony tool for the Model Context Protocol (MCP) that lets AI agents make real outbound phone calls, retrieve conversation transcripts, and check call status. It is a proprietary cloud service hosted by Stexa.

How to use Stexa Voice MCP?

Register at stexa.ru, create an API key (format stx_live_…) in the Cabinet’s MCP Server section, then configure your MCP client (Claude Desktop, Cursor, Python, n8n) with the endpoint https://stexa.ru/mcp/v1/jsonrpc and a Bearer token in the Authorization header. OAuth 2.1 is recommended for production – set "auth": "oauth2" in the client configuration.

Key features of Stexa Voice MCP

• Make outbound voice calls with the AI bot Alina (stexa_call_outbound)
• Check call status and get live updates (stexa_get_call_status)
• Retrieve full or partial transcripts (stexa_get_transcript)
• List recent calls with filters (stexa_list_recent_calls)
• OAuth 2.1 (Authorization Code + PKCE) support
• HTTP JSON-RPC 2.0 transport with rate limits

Use cases of Stexa Voice MCP

• AI-triggered outbound campaigns (e.g., CRM events)
• Appointment reminders delivered via voice
• Post-call feedback and NPS collection
• Enriching CRM with real call transcripts
• Voice bot booking for salons, clinics, and auto-services

FAQ from Stexa Voice MCP

How do I authenticate – API key or OAuth?

Both are supported. Create an API key (stx_live_…) in the Cabinet. For production, use OAuth 2.1 by setting "auth": "oauth2" in your MCP client config; the server uses Authorization Code + PKCE via an external browser flow.

What are the rate limits?

Per IP: 60 requests/minute. Per API key: 120 requests/minute. The stexa_call_outbound tool is limited to 10 calls/minute per key.

What scopes are available and which tools do they grant?

call:write grants stexa_call_outbound; call:read grants stexa_get_call_status and stexa_list_recent_calls; transcript:read grants stexa_get_transcript. API keys from the Cabinet include all three scopes by default.

Where does the call data live and what are the limits?

Stexa Voice MCP is a cloud service. Call duration can be set between 30–600 seconds (default 300). The wait_for_completion parameter blocks up to 90 seconds for a synchronous transcript. Transcripts and summaries are stored on Stexa’s servers – see their privacy policy.

What transport and protocol does it use?

HTTP JSON-RPC 2.0, protocol version 2024-11-05. The endpoint is POST https://stexa.ru/mcp/v1/jsonrpc. Tools can be discovered without authentication via GET to the same URL.