elevenlabs-mcp
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| FFMPEG_PATH | No | Path to the ffmpeg binary, required for joining long dialogue and assembling episodes. | |
| FFPROBE_PATH | No | Path to the ffprobe binary. | |
| ELEVENLABS_API_KEY | No | Your ElevenLabs API key (e.g. sk_...). Overrides the credentials file. | |
| ELEVENLABS_BASE_URL | No | ElevenLabs API base URL, e.g. https://api.eu.residency.elevenlabs.io. | |
| ELEVENLABS_OUTPUT_DIR | No | Directory where generated audio files are saved locally. Default: ~/elevenlabs-output. | |
| ELEVENLABS_CREDENTIALS | No | Path to the credentials env file containing ELEVENLABS_API_KEY=sk_... (default ~/.config/elevenlabs-mcp/credentials.env). |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": true
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| check_subscriptionB | Plan, credits used and left, and when the quota resets. |
| list_modelsA | Speech models with their languages. eleven_v3 is the most expressive (dialogue, audio tags); eleven_multilingual_v2 is the most stable for long narration; flash/turbo are fast and cheap. |
| list_voicesC | Voices in your own library (premade, cloned, designed and added from the shared library), with their voice_id. |
| search_voice_libraryA | Search ElevenLabs' shared voice library (thousands of community voices, by language/accent/gender/use case). Add a voice with add_library_voice before using it. |
| add_library_voiceB | Add a shared-library voice to your own voices so it can be used for speech. Uses a voice slot. |
| text_to_speechB | One voice reading a text — narration, a cold open, an ad read. For several speakers use text_to_dialogue. |
| text_to_dialogueA | Multi-speaker conversation in one natural take (eleven_v3) — the core podcast tool. Pass the script as ordered lines, each with a voice_id. Long scripts are split into ~2 000-character requests (whole lines kept together) and joined; locally that needs ffmpeg. Audio tags in the text steer delivery: [laughs], [sighs], [whispers], [excited], [interrupting]. |
| compose_musicB | Original music from a text prompt — intro/outro themes, beds under speech, transitions. Licensed for commercial use on paid plans. |
| sound_effectC | A sound effect from a description — stingers, whooshes, ambience, applause. |
| list_outputsB | Audio files in the output directory, newest first, with duration. |
| assemble_episodeA | Join audio files from the output directory into a finished episode, in order: e.g. intro music → dialogue → stinger → dialogue → outro. Per segment: fades, a gap after, gain. The result is normalised to -16 LUFS (podcast standard), MP3 192 kbps. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 11 tools
Each tool targets a distinct resource or operation: subscription, models, own voices, shared library search/add, single-voice TTS, multi-voice dialogue, music, SFX, assembly, and outputs. The only potential overlap (text_to_speech vs text_to_dialogue) is explicitly clarified by the descriptions.
Names are consistently snake_case and mostly follow a verb_noun pattern (list_models, text_to_speech, compose_music, add_library_voice). The one minor deviation is sound_effect, which is noun-only rather than verb-led, but it remains predictable.
11 tools is well-scoped for an audio production server. Each tool covers a discrete stage from subscription checking and voice discovery through generation, assembly, and output listing.
Core podcast/audio workflow is covered: quota check, model/voice discovery, library addition, TTS/dialogue/music/SFX generation, episode assembly, and output listing. Missing voice cloning/design operations (implied by list_voices description) and delete/update voice or output management, but these are minor for the primary use case.