Suno Brings AI Music Creation Directly Into ChatGPT, Claude and Gemini
Suno's new MCP server lets ChatGPT, Claude, Gemini, and Cursor generate songs and looping visuals directly from chat without any API keys.
Takeaways
- Suno MCP is live for all plans, connecting Suno to ChatGPT, Claude, Gemini, Cursor, and Claude Code.
- Endpoint is
https://api.suno.com/mcp, OAuth sign-in, no API key required. - Supports song creation, voice-memo-to-song, covers, mashups, and extensions from chat.
- Looping visuals and cover art generation available on paid plans only.
- Uses existing Suno credits; outputs save automatically to your Suno library.
- Full setup guide and FAQ on the Suno help center.
Suno has launched a hosted Suno MCP server, giving compatible AI clients such as ChatGPT, Claude, Gemini, Claude Code, and Cursor access to its music-generation tools. From the conversation, an assistant can generate songs, transform existing tracks, and create looping visuals.
The Model Context Protocol is an open standard that lets AI clients discover and call external tools. With Suno connected, the assistant can pass lyrics, creative briefs, style notes, and other conversation context directly into a generation request. The resulting work appears in the user’s regular Suno library.
From chat context to tracks
The server exposes much of Suno’s existing creation workflow through MCP tool calls:
- Generate songs from plain-language descriptions.
- Add lyrics and musical styles before rendering a track.
- Turn voice memos or soundbites into fuller arrangements.
- Create covers, remixes, mashups, and extensions from library tracks.
- Generate artwork and looping visuals for songs.
Artwork and looping visuals require a paid Suno plan. Song generation is available across Suno plans, including Free, and consumes the credits included with the connected account.
Connect through one endpoint
The hosted server runs at https://api.suno.com/mcp. Authentication uses Suno’s OAuth flow, so the client opens a sign-in and authorization step after connecting to the endpoint.
Claude Code can add the server with one command:
claude mcp add --transport http suno https://api.suno.com/mcp
Cursor reads the connection from mcp.json:
{
"mcpServers": {
"suno": {
"url": "https://api.suno.com/mcp"
}
}
}
ChatGPT, Claude, and Gemini provide custom-connector interfaces where available. Add a connector named Suno, enter the endpoint URL, and complete OAuth authorization. Menu names and connector availability vary by client and subscription tier.
Prompts that keep their context
Passing the active conversation into Suno removes the need to restate a project’s lyrics, tone, structure, or production brief in a separate music interface. Useful workflows include:
- Draft lyrics in Claude, then generate a track with a specified genre, tempo, and vocal style.
- Attach a voice memo in a supported client and request an indie-folk arrangement around its melody.
- Create a lo-fi track from Cursor, then revise it with follow-up instructions such as “reduce the tempo” or “add a bridge.”
- Generate an eight-second looping visual for a song and social clip within the same conversation.
- Use a video brief already in the thread to generate background music that matches its pacing and mood.
Where the connector stops
The MCP server calls Suno’s existing models, so output quality, supported styles, voice restrictions, and transformation capabilities match the company’s web product. Repeated generations also consume credits at the normal rate. Automated agent loops can exhaust an allowance quickly, making bounded retries and tool-call approvals useful safeguards.
Client support creates a separate access constraint. Some assistants reserve custom MCP connectors for particular paid tiers, and ChatGPT’s custom MCP support is not available on every plan. A user may therefore have sufficient Suno access while lacking connector support in the chosen assistant.
File-based workflows also depend on whether the client can pass attachments or audio references to the connector. Text prompts have the broadest compatibility, while voice memos and soundbites may require client-specific handling.
Audio joins the agent stack
A standard MCP client and a Suno login now provide an audio-generation step that agents can invoke alongside search, code execution, and image generation. Teams building automated video, podcast, game, or social-content pipelines can route existing project context into Suno without maintaining a separate Suno-specific integration.