# mcp-happyhorse [Health: Active]

**Category:** 🎥 Multimedia Process  
**Repository:** https://github.com/AceDataCloud/HappyHorseMCP  
**GitHub Stars:** 0  
**Views:** 0  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/mcp-happyhorse

## Description
Generate and edit Happy Horse AI videos through Ace Data Cloud

## Tools
Capabilities this server exposes over MCP:

- **happyhorse_generate_video** — Generate a video from text
- **happyhorse_generate_video_from_image** — Animate one first-frame image
- **happyhorse_generate_video_from_references** — Generate from 1-9 reference images
- **happyhorse_edit_video** — Edit a source video with up to 5 references
- **happyhorse_get_task** — Query one task
- **happyhorse_get_tasks_batch** — Query multiple tasks
- **happyhorse_list_models** — List valid models for each action

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `uvx` (confidence: high):

```json
"mcpServers": {
  "mcp-happyhorse": {
    "command": "uvx",
    "args": ["mcp-happyhorse"],
    "env": {
      "ACEDATACLOUD_API_TOKEN": ""
    }
  }
}
```

**Requires environment variables:** `ACEDATACLOUD_API_TOKEN` — the values above are empty placeholders; fill in real credentials before running (see the repository for what each one is for).

## Documentation

## What mcp-happyhorse does

The mcp-happyhorse MCP server exposes Ace Data Cloud's Happy Horse AI video operations through Model Context Protocol. It belongs to the Multimedia Process category and is intended for clients that need to submit video generation or editing tasks, then inspect their progress.

Generation can start from a text prompt, one image used as the first frame, or between one and nine reference images describing subjects or style. Editing accepts a source video and up to five reference images. The service supports 720P and 1080P output, subject to the selected operation and model.

## How it works

The mcp-happyhorse MCP server forwards tool requests to the Ace Data Cloud API. Video operations create tasks rather than returning an immediate finished result. Use the single-task lookup tool for one task or the batch lookup tool for multiple tasks. The model-listing tool provides the valid models available for each action.

The server can be reached through the hosted Streamable HTTP endpoint at `https://happyhorse.mcp.acedata.cloud/mcp`, or run locally with stdio. Hosted connections support Ace Data Cloud OAuth or an explicit Bearer token. A local process uses the Ace Data Cloud API and requires an API token in its environment.

Available tools include:

- `happyhorse_generate_video` for text-to-video generation.
- `happyhorse_generate_video_from_image` for animating one first-frame image.
- `happyhorse_generate_video_from_references` for generation from 1-9 reference images.
- `happyhorse_edit_video` for editing a source video with up to five references.
- `happyhorse_get_task` for checking one task.
- `happyhorse_get_tasks_batch` for checking multiple tasks.
- `happyhorse_list_models` for listing valid models by action.

## Setup and configuration

For hosted use, add the MCP URL to a client that supports remote MCP and complete Ace Data Cloud sign-in when prompted. OAuth does not require pasting a token into the client configuration. Alternatively, send an Ace Data Cloud API token as an `Authorization: Bearer` header.

For local stdio, install the `mcp-happyhorse` package and provide `ACEDATACLOUD_API_TOKEN` to the process. Keep the token in a secret store or environment variable rather than placing it in prompts, source control, screenshots, or shared workflows. Do not configure both OAuth and a fixed authorization header for the same connection.

## Limitations and notes

The mcp-happyhorse MCP server depends on Ace Data Cloud account access, service or model permissions, available balance, and any applicable usage limits or pricing. A task-lookup permission does not necessarily authorize submitting a generation request. Review the current Ace Data Cloud service documentation before running paid operations.

Reference-image limits differ by operation: reference-to-video accepts one to nine images, while video editing accepts up to five. The first-frame workflow is separate and uses one image. Client configuration syntax also varies between hosted OAuth, Bearer-token connections, and local stdio, so credentials should be configured according to the selected transport.

_Full upstream README: https://allmcps.com/mcp/mcp-happyhorse/readme_