Skip to main content
Glama
iceener

ElevenLabs MCP Server

by iceener
OverviewSchemaSourceRelated ServersDiscussionsAdmin
TypeScript
Hybrid

ElevenLabs MCP Server

Fetch-native MCP server for ElevenLabs text-to-speech, speech-to-text, voices, and transcripts. It runs on Bun and Cloudflare Workers.

This repository targets the prerelease 2026-07-28 MCP candidate with @modelcontextprotocol/server@2.0.0-beta.5. It does not claim final conformance until the dated specification and stable packages are published and verified.

Tools

  • health

  • text_to_speech — returns generated audio as a base64 data-URI resource

  • speech_to_text — accepts an HTTPS URL or base64 file content

  • list_voices

  • get_voice

  • get_transcript

  • delete_transcript

Tool names, inputs, annotations, audio behavior, and transcript behavior are shared across Bun and Workers. Tools that advertise output schemas return matching structuredContent.

Related MCP server: ElevenLabs MCP Server

Configure

bun install
cp .env.example .env

Set the provider credential:

ELEVENLABS_API_KEY=xi_your_elevenlabs_key

This is the only credential sent to api.elevenlabs.io.

Optional MCP caller authentication is separate:

AUTH_STRATEGY=api_key
API_KEY=your_mcp_caller_key
API_KEY_HEADER=x-api-key

API_KEY, BEARER_TOKEN, custom caller headers, and OAuth resource tokens are never reused as ElevenLabs credentials. External JWT Resource Server mode is documented in .env.example.

Run

bun run dev
# http://127.0.0.1:3000/mcp

Cloudflare Workers:

bun x wrangler secret put ELEVENLABS_API_KEY --config wrangler.jsonc
bun run deploy

Set a real MCP_PUBLIC_URL, Host/Origin allowlists, and binding IDs in wrangler.jsonc before deployment.

Protocol and HTTP behavior

  • One deployment-scoped fetch handler and a fresh McpServer per request

  • Modern protocol pinned to candidate 2026-07-28

  • SDK stateless fallback for legacy 2025-11-25 clients

  • GET /mcp and DELETE /mcp return 405; no session capability is advertised

  • Bounded MCP request bodies, strict Host/Origin checks, and allowlisted CORS headers

  • Cancellation reaches ElevenLabs fetches through ctx.mcpReq.signal

Endpoints:

Endpoint

Methods

Purpose

/mcp

POST, OPTIONS

MCP and CORS preflight

/health

GET

Runtime health

/icon.svg

GET

Server icon

Validate

bun test
bun run typecheck
bun run lint
bun run format:check
bun run build
bun run build:worker
bun run types:worker:check

Provider tests use mocks; real speech generation and transcription require a valid ElevenLabs API key and may incur usage charges.

License

MIT

This server cannot be deployed

One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

Maintenance

ActivitySlowing
ResponsivenessNo issues

Resources

Related MCP Connectors

Related MCP Servers

  • ElevenLabs MCP Server by elevenlabs

    ElevenLabs MCP Serverofficial

    elevenlabs
    A
    license
    B
    quality
    F
    maintenance
    An official Model Context Protocol (MCP) server that enables AI clients to interact with ElevenLabs' Text to Speech and audio processing APIs, allowing for speech generation, voice cloning, audio transcription, and other audio-related tasks.
    27
    1,539
    MIT

  • Edge TTS MCP

    Hwenyi
    A
    license
    Not graded
    quality
    D
    maintenance
    An MCP server that converts text into lifelike speech using Microsoft Edge's Text-to-Speech service, supporting customizable voice, rate, volume, and pitch.
    4
    MIT