Ghibli Video MCP Server
# mcp-server-ghibli MCP Server
[](https://smithery.ai/server/@MichaelYangjson/mcp-ghibli-video)
A TypeScript-based MCP server that provides AI image and video generation capabilities through a simple interface.
> **Note**: This server requires an API key from [GPT4O Image Generator](https://www.gpt4oimg.com/). Please visit the website to obtain your API key before using this service.
## Features
### Tools
#### 1. Image to Video Conversion
- `image_to_video` - Convert static images into animated videos
- Required parameters:
- `image`: Base64 encoded image or image URL
- `api_key`: Authentication key
- Optional parameters:
- `prompt`: Text prompt to guide video generation (default: "in the style of ghibli")
- `aspect_ratio`: Output video aspect ratio (default: "9:16")
- `negative_prompt`: Negative prompt to guide generation (default: "bad prompt")
#### 2. Points Management
- `get_points` - Check remaining API credits
- Required parameters:
- `api_key`: Authentication key
#### 3. Task Management
- `get_task_result` - Check the status of a video generation task
- Required parameters:
- `taskId`: Task ID returned from image_to_video
- `api_key`: Authentication key
## Development
Install dependencies:
```bash
npm install
```
Build the server:
```bash
npm run build
```
For development with auto-rebuild:
```bash
npm run watch
```
## Installation
To use with Claude Desktop, add the server config:
On MacOS: `~/Library/Application Support/Claude/claude_desktop_config.json`
On Windows: `%APPDATA%/Claude/claude_desktop_config.json`
```json
{
"mcpServers": {
"mcp-server-ghibli-video": {
"command": "npx",
"args": ["-y", "@openmcprouter/mcp-server-ghibli-video"],
"env": {
"Ghibli_API_URL": "https://www.gpt4oimg.com"
}
}
}
}
```
### Installing via Smithery
To install mcp-server-ghibli MCP Server for Claude Desktop automatically via [Smithery](https://smithery.ai/server/@MichaelYangjson/mcp-ghibli-video):
```bash
npx -y @smithery/cli install @MichaelYangjson/mcp-ghibli-video --client claude
```
### Debugging
Since MCP servers communicate over stdio, debugging can be challenging. We recommend using the [MCP Inspector](https://github.com/modelcontextprotocol/inspector), which is available as a package script:
```bash
npm run inspector
```
The Inspector will provide a URL to access debugging tools in your browser.
TDQS
Scored across 3 tools
Each tool has a clearly distinct purpose with no overlap: get_points retrieves points, get_task_result fetches task results, and image_to_video handles video conversion. An agent can easily differentiate them based on their specific functions.
The tools follow a consistent verb_noun pattern (get_points, get_task_result, image_to_video), with minor deviations in verb usage (get vs. image_to). The naming is readable and mostly predictable, though not perfectly uniform.
With only 3 tools, the set feels thin for a video server's apparent scope, lacking operations like video management, task creation, or point usage. This minimal count may limit agent workflows and indicate incomplete coverage.
There are significant gaps in the tool surface for a video server domain: no tools for creating tasks, managing videos, or using points, and core operations like video playback or editing are missing. This will likely cause agent failures in handling typical video-related workflows.