mcp server
ccapi-mcp
CCAPI MCP server — image, video, music and text generation across 100+ models through one endpoint
Set up
Runs on your computer. Pick your agent and operating system.
This project does not publish an install method we can turn into a config. Follow the README instead.
README
synced Oct 23, 2026CCAPI MCP Server
Image, video, music and text generation across 100+ models, over the Model Context Protocol — one hosted endpoint, no local process to run.
Endpoint: https://api.ccapi.ai/mcp (Streamable HTTP)
Docs: https://ccapi.ai/mcp
Registry: ai.ccapi/mcp ·
Glama
This repository holds the published server manifest and the client documentation. The gateway implementation itself is closed source.
Install
{
"mcpServers": {
"ccapi": {
"type": "http",
"url": "https://api.ccapi.ai/mcp",
"headers": { "Authorization": "Bearer sk-YOUR-KEY" }
}
}
}
Claude Code:
claude mcp add --transport http ccapi https://api.ccapi.ai/mcp \
--header "Authorization: Bearer sk-YOUR-KEY"
A CCAPI API key is required — the server returns 401 without one. Keys are
issued at https://ccapi.ai. list_models, get_balance and get_task do not
consume quota; the generation tools do.
Tools
Eleven tools across five jobs.
| Tool | What it does |
|---|---|
generate_image | Text to image |
edit_image | Edit an existing image from a prompt |
upscale_image | Increase resolution |
generate_video | Text or image to video, optionally with a soundtrack |
generate_music | Prompt to a full track |
generate_lyrics | Write lyrics |
extend_music | Continue an existing track |
chat_completion | Text generation across the available models |
list_models | Resolve model names and see what is available |
get_balance | Remaining quota |
get_task | Poll an asynchronous job |
Video and music generation are asynchronous: those tools return a task id, and
the assistant polls get_task until the job finishes.
Scoped endpoints
Alongside the full endpoint, four scoped endpoints each carry a single capability plus the shared discovery and polling tools:
| Endpoint | Tools |
|---|---|
https://api.ccapi.ai/mcp | all eleven |
https://api.ccapi.ai/mcp/image | generate_image, edit_image, upscale_image + shared |
https://api.ccapi.ai/mcp/video | generate_video + shared |
https://api.ccapi.ai/mcp/music | generate_music, generate_lyrics, extend_music + shared |
https://api.ccapi.ai/mcp/text | chat_completion + shared |
Shared across every scope: list_models, get_balance, get_task.
They exist because tool selection degrades as the list grows — a model choosing
between four tools picks correctly more often than one choosing between eleven.
If an agent only ever generates images, point it at /mcp/image.
The registry advertises only https://api.ccapi.ai/mcp, since a client should
be offered one server rather than five variants of the same one. The scoped
endpoints are a deliberate choice by the integrator, not a menu.
Manifest
server.json mirrors what is published to the official MCP
registry as ai.ccapi/mcp. Ownership is proved by domain verification against
https://ccapi.ai/.well-known/mcp-registry-auth.
Licence
MIT, covering the documentation and manifest in this repository. Use of the hosted service is governed by the terms at https://ccapi.ai.