# AI or Not MCP Server

[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
[![npm version](https://img.shields.io/npm/v/aiornot-mcp.svg)](https://www.npmjs.com/package/aiornot-mcp)
[![MCP](https://img.shields.io/badge/MCP-Compatible-blue)](https://modelcontextprotocol.io)

An MCP (Model Context Protocol) server that integrates with the [AI or Not](https://aiornot.com) API to detect AI-generated content in images, videos, audio, and text.

## Features

- **Image Analysis**: Detect AI-generated images, deepfakes, NSFW content, and image quality issues
- **Video Analysis**: Detect AI-generated video, synthetic voices, AI music, and video deepfakes
- **Audio Analysis**: Detect AI-generated music and synthetic voices
- **Text Analysis**: Detect AI-written text with confidence scoring and annotations
- **API Health Check**: Verify API availability

## Prerequisites

- Node.js 18+
- An API key from [AI or Not](https://aiornot.com)

## Installation

### From Source

```bash
git clone https://github.com/tymrtn/aiornot-mcp.git
cd aiornot-mcp
npm install
npm run build
```

### From npm (coming soon)

```bash
npm install -g aiornot-mcp
```

## Configuration

### Environment Variables

| Variable | Required | Default | Description |
|----------|----------|---------|-------------|
| `AIORNOT_API_KEY` | Yes | - | Your AI or Not API key |
| `AIORNOT_API_URL` | No | `https://api.aiornot.com` | API base URL |

### Claude Desktop / Claude Code

Add to your MCP settings file:

```json
{
  "mcpServers": {
    "aiornot": {
      "command": "node",
      "args": ["/path/to/aiornot-mcp/build/index.js"],
      "env": {
        "AIORNOT_API_KEY": "your_api_key_here"
      }
    }
  }
}
```

**Settings file locations:**
- Claude Desktop (macOS): `~/Library/Application Support/Claude/claude_desktop_config.json`
- Claude Code: `~/.claude/mcp_servers.json`

## Usage

### Running the Server

```bash
AIORNOT_API_KEY="your_api_key_here" node build/index.js
```

### Available Tools

#### `aiornot_analyze_media`

Analyze media content for AI generation.

**Parameters:**

| Parameter | Type | Required | Description |
|-----------|------|----------|-------------|
| `media_type` | string | Yes | One of: `image`, `video`, `text`, `audio_music`, `audio_voice` |
| `file_path` | string | Conditional | Path to file (required for image/video/audio) |
| `text` | string | Conditional | Text content (required for text, min 250 chars) |
| `only` | string[] | No | Report types to include |
| `excluding` | string[] | No | Report types to exclude |
| `external_id` | string | No | Tracking identifier |
| `include_annotations` | boolean | No | Include block-level annotations (text only) |
| `timeout_ms` | number | No | Override request timeout |

**Report Types by Media:**

| Media Type | Available Reports |
|------------|-------------------|
| Image | `ai_generated`, `deepfake`, `nsfw`, `quality`, `reverse_search` |
| Video | `ai_video`, `ai_music`, `ai_voice`, `deepfake_video` (off by default) |

#### `aiornot_is_live`

Check if the AI or Not API is available.

### Examples

**Analyze an image:**
```json
{
  "media_type": "image",
  "file_path": "/path/to/image.jpg"
}
```

**Analyze an image for specific checks:**
```json
{
  "media_type": "image",
  "file_path": "/path/to/image.jpg",
  "only": ["ai_generated", "deepfake"]
}
```

**Analyze video including deepfake detection:**
```json
{
  "media_type": "video",
  "file_path": "/path/to/video.mp4",
  "only": ["ai_video", "deepfake_video"]
}
```

**Analyze text:**
```json
{
  "media_type": "text",
  "text": "Your text content here (minimum 250 characters)...",
  "include_annotations": true
}
```

**Analyze audio for synthetic voice:**
```json
{
  "media_type": "audio_voice",
  "file_path": "/path/to/audio.mp3"
}
```

### Response Format

The server returns structured JSON with:
- `media_type`: The analyzed media type
- `scores`: Extracted confidence scores and verdicts
- `response`: Full API response

Example response for image analysis:
```json
{
  "media_type": "image",
  "scores": {
    "ai_generated": {
      "verdict": "ai",
      "ai_confidence": 0.95,
      "human_confidence": 0.05
    },
    "deepfake": {
      "is_detected": false,
      "confidence": 0.02
    }
  },
  "response": { ... }
}
```

## Timeouts

Default timeouts vary by media type:

| Media Type | Default Timeout |
|------------|-----------------|
| Image | 30 seconds |
| Text | 30 seconds |
| Video | 120 seconds |
| Audio (music) | 120 seconds |
| Audio (voice) | 120 seconds |

Use `timeout_ms` to override if needed.

## Development

```bash
# Install dependencies
npm install

# Build
npm run build

# Watch mode
npm run watch

# Test with MCP Inspector
npm run inspector
```

## License

MIT - see [LICENSE](LICENSE)

## Links

- [AI or Not](https://aiornot.com) - API provider
- [Model Context Protocol](https://modelcontextprotocol.io) - MCP specification
- [MCP SDK](https://github.com/modelcontextprotocol/typescript-sdk) - TypeScript SDK