Model Context Protocol

Your voice agents, wired into every AI tool

MCP is an open standard that lets AI apps and tools talk to each other in a common language — one adapter instead of a bespoke integration per pairing. Vomyra speaks it in both directions: Claude, ChatGPT and Perplexity can run your voice agents, and your agents can use your tools mid-call.

Server endpoint https://console.vomyra.com/mcp · Streamable HTTP · OAuth 2.1

How it works

MCP runs in both directions

Vomyra is an MCP server and an MCP client — one protocol doing two jobs. Your AI tools drive your agents, and your agents reach your systems.

Vomyra as an MCP server

Run Vomyra from your AI tool

01

Point Claude, ChatGPT or Perplexity at https://console.vomyra.com/mcp and they can build assistants, place calls and read transcripts for you — in plain English, without opening the dashboard.

  • Sign in once with OAuth
  • 24 actions across your account
  • No code, no SDK
MCP server docs

Vomyra as an MCP client

Let your agent use your tools mid-call

02

Attach your own MCP servers to an assistant and it looks up live data and takes real actions while the caller is still on the line — order status, availability, customer history.

  • Added from the dashboard, no code
  • Whitelist exactly which tools it may use
  • Streamable HTTP or SSE
MCP client docs
What you can do

24 actions, four areas

Once connected, your AI tool reaches the same operations you'd perform in the dashboard — driven by conversation.

Assistants10

Create, configure and manage voice assistants and their settings.

Tools6

Build and control the actions your assistants can perform on a call.

Calls5

Place outbound calls and pull conversation transcripts.

Numbers3

Inspect phone numbers and how calls route to your assistants.

Endpoint
https://console.vomyra.com/mcp
Transport
Streamable HTTP (JSON-RPC 2.0)
Server name
vomyra
Version
1.0.0
Auth
OAuth 2.1 or API key
Rate limit
600 req/min authenticated

OAuth tokens refresh hourly; refresh tokens last 30 days.

API keys are for scripts and backend automation.

30 req/min unauthenticated, 600 authenticated.

FAQ

MCP questions

What is MCP?

MCP (Model Context Protocol) is an open standard that lets AI apps and tools talk to each other in a common language. Think of it as a universal adapter: instead of writing a bespoke integration for every pairing, a tool exposes an MCP server once and any MCP-speaking app can use it.

Which AI tools can connect to Vomyra over MCP?

Any MCP client that supports remote servers over Streamable HTTP. We publish step-by-step guides for Claude and ChatGPT, and Vomyra also works with Perplexity's custom remote connectors and other MCP-speaking tools.

What can an AI tool actually do with Vomyra over MCP?

Vomyra's MCP server exposes 24 actions across four areas: assistants (10), tools (6), calls (5) and numbers (3). That covers creating and configuring assistants, building the actions they can take, placing outbound calls, reading transcripts and inspecting number routing.

How does Vomyra's MCP server authenticate?

Two ways. OAuth 2.1 is the sign-in flow AI tools use — access tokens refresh automatically every hour and refresh tokens last 30 days. API keys, generated in the dashboard, are for scripts and backend automation.

Can Vomyra assistants use my MCP servers during a call?

Yes — that's the other direction. As an MCP client, Vomyra can attach your MCP servers to an assistant so it looks up live data and takes real actions mid-call. You add the server in the dashboard, whitelist the tools the assistant may use, and attach it. It supports Streamable HTTP and SSE, with bearer token, custom header or OAuth 2 refresh token auth.

Do I need to write code to connect an MCP server to my assistant?

No. MCP servers are added and attached from the Vomyra dashboard. You enter the server URL and auth, test the connection to discover its tools, whitelist the ones you want, and attach it to an assistant. Only calls started after attaching will see the new tools.