Add the server
Point your MCP client at mcp.apiframe.ai with your API key in the X-API-Key header. New accounts start with free credits.
Model Context Protocol
Give your AI agent 70+ media models.
The Apiframe MCP server lets Claude, Cursor, ChatGPT, and any MCP-compatible client generate images, video, and music mid-conversation. Same catalog, same credits, same API key as the REST API.
Remote MCP server
https://mcp.apiframe.ai
- Auth
- X-API-Key header, same key as the REST API
- Transport
- Streamable HTTP
Works with
Claude Cursor ChatGPT Windsurf Agents SDK LangGraph
Setup
Two minutes from key to first generation.
Add the server to your client config, paste your API key, and ask your agent for an image. No SDK, no local process to babysit. The server runs on our side.
~/.cursor/mcp.json
{
"mcpServers": {
"apiframe": {
"url": "https://mcp.apiframe.ai",
"headers": {
"X-API-Key": "YOUR_API_KEY"
}
}
}
}Your key stays in your client config and is sent per request. The server holds no session state. Create or rotate keys anytime in the dashboard.
Tools
Every generation endpoint, exposed as a tool.
The server mirrors the REST API: the same models, parameters, and credit rates, wrapped as MCP tools your agent can discover and call.
generate_image
Text-to-image and image-to-image across Midjourney, Flux, Nano Banana, GPT Image, and every other image model.
generate_video
Text- and image-to-video with Seedance, Kling, Veo, Sora, Hailuo, and the rest of the video catalog.
generate_music
Full songs and instrumentals with Suno, Udio, Lyria, Mureka, and more.
edit_image
Inpainting and prompt-based edits with Flux Fill Pro and friends.
upscale_image
Topaz and Clarity upscalers for print-ready resolution.
upscale_video
Topaz Video upscale to 4K.
remove_background
Clean cutouts with Bria and 851-labs background removal.
get_job
Poll any job for status and CDN result URLs.
list_models
The live model catalog with pricing and capabilities, so agents can pick the right model themselves.
How it works
Your agent prompts, Apiframe renders.
Ask in plain language
“Generate a 10-second product teaser with Seedance.” The agent picks the tool, fills the parameters, and submits the job.
Get CDN links back
The tool returns job status and hosted result URLs your agent can embed, download, or hand to the next step in its workflow.
Use cases
What teams wire up with MCP.
Agentic creative pipelines
Agents that draft copy, generate matching visuals, and assemble ad variants end to end without leaving the conversation.
Media inside coding assistants
Generate placeholder-free hero images, icons, and demo videos for your app directly from Cursor or Claude Code.
Chat products with generation
Ship image and music generation in your own assistant by pointing it at one MCP server instead of five provider APIs.
Automated content ops
Scheduled agents that turn briefs into rendered assets (thumbnails, jingles, teasers) and file them where they belong.
MCP questions
Everything about the server, auth, and billing. Anything else, ask us on Discord or through the contact page.
What is the Model Context Protocol?
MCP is an open standard that lets AI assistants and agents call external tools. A client like Claude or Cursor connects to an MCP server, discovers its tools, and invokes them during a conversation. The Apiframe MCP server exposes the full generation API as tools, so any MCP client can create images, video, and music.
How do I authenticate?
With your regular Apiframe API key, sent as an X-API-Key header, exactly like the REST API. Keys are created and rotated in the dashboard, and the same team, credits, and permissions apply.
How is usage billed?
Identically to the REST API. Each tool call spends credits at the model's normal rate, failed jobs are refunded automatically, and everything shows up in the same usage analytics.
Which clients are supported?
Any client that speaks MCP over streamable HTTP: Claude Desktop and Claude Code, Cursor, ChatGPT, Windsurf, and agent frameworks like the OpenAI Agents SDK or LangGraph. If your client only supports stdio servers, bridge it with mcp-remote.
Is the MCP server rate-limited differently?
No. Tool calls count against the same per-plan rate and concurrency limits as REST requests. Your plan's concurrent-job cap applies to jobs started over MCP too.
Can my agent pick the model on its own?
Yes. The list_models tool returns the live catalog with capabilities and per-model pricing, so an agent can choose a model that fits the brief and budget. Or you can pin one in your prompt.
9,800+
developers & companies
4.8 on G2
Official partner of
Power your next AI product with Apiframe.
Instant access to 70+ media models through a single API. Start free and scale when you're ready.