MCP for AI agents
Let your AI agent turn text into audio
Voxi.fm ships an MCP server, so agents like Claude Code, Claude Desktop and Cursor can generate natural-sounding audio directly — no browser, no copy-paste. Ask your agent to narrate a draft and the MP3 lands in your Voxi.fm library.
Get a free API keyWhat is MCP?
The Model Context Protocol is an open standard for connecting AI agents to external tools. A Voxi.fm API key plus one config line gives your agent a set of audio tools it can call on its own — the agent decides when to list voices, start a generation, and check whether it finished.
What agents can do
generate_audioConvert text into audio with a chosen voice, as a new article or a new version of one (costs 1 credit)get_audio_statusPoll a generation until it completes and get the MP3 URL and embed codelist_voicesList available voice profileslist_audioList recent audio items in your librarylist_articlesList your articles, optionally only one project'sget_articleGet an article's project, audio versions and embed codeset_article_projectMove an article to another project, or out of its projectdelete_articleDelete an article with all its audio, e.g. when its page is taken downlist_projectsList your projects and their embed player defaultscreate_projectCreate a project, optionally with its embed player defaultsupdate_projectChange a project's settings, e.g. its embed theme and designget_creditsCheck your remaining credits
Connect in one line
Claude Code:
claude mcp add --transport http voxi https://us-central1-voxi-6c6fa.cloudfunctions.net/mcpServer --header "Authorization: Bearer YOUR_API_KEY"
Claude Desktop, Cursor, and other MCP clients:
{
"mcpServers": {
"voxi": {
"type": "http",
"url": "https://us-central1-voxi-6c6fa.cloudfunctions.net/mcpServer",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}The endpoint is https://us-central1-voxi-6c6fa.cloudfunctions.net/mcpServer. Create your key on the Agent Access page after signing in, then swap it in for YOUR_API_KEY.
Who it's for
- Writers whose drafting already happens inside an AI agent
- Publishing pipelines that generate audio for every new post
- Teams building custom agents on the MCP SDKs
- Anyone who would rather ask than click through a wizard
Same credits, no extra fee
One credit per article, whether it comes from the app or an agent. Five free articles to start, no credit card required.
See pricingFAQ
Frequently asked questions
What is MCP?
The Model Context Protocol is an open standard for connecting AI agents to external tools. Once an agent speaks MCP, it can call any MCP server — including Voxi.fm — without custom integration code.
Which agents can use the Voxi.fm MCP server?
Any MCP client that supports the Streamable HTTP transport, including Claude Code, Claude Desktop, Cursor, and custom agents built on the MCP SDKs.
What does it cost?
The same as generating audio in the app: 1 credit per article. New accounts get 5 free credits, and paid plans start low. There is no extra charge for agent access.
How long does a generation take?
Generation is asynchronous. generate_audio returns an audio id immediately; the agent polls get_audio_status until the MP3 URL appears, typically after 1–5 minutes depending on article length.
Is it safe to give an agent an API key?
A key only works on your own account. It can create, read and delete your articles and projects, but it can’t change billing or reach anyone else’s data. Keys are stored hashed, never in plaintext, and you can revoke one at any time from the Agent Access page.
Where does the generated audio end up?
In your normal Voxi.fm library, alongside audio created in the app. You get the same MP3 and embed player.
Give your agent a voice
Create an account, mint an API key, and connect in under a minute.
Create accountAlready signed in? Manage your API keys.