multimodal-mcp
Multi-provider media generation MCP server that generates images, videos, audio, and transcriptions from text prompts using OpenAI, xAI, Gemini, ElevenLabs, and BFL through a single unified interface.
README
multimodal-mcp
Multi-provider media generation MCP server. Generate images, videos, audio, and transcriptions from text prompts using OpenAI, xAI, Gemini, ElevenLabs, and BFL (FLUX) through a single unified interface.
Features
- 🎨 Image Generation — Generate images via OpenAI (gpt-image-1), xAI (grok-imagine-image), Gemini (imagen-4), or BFL (FLUX Pro 1.1)
- ✏️ Image Editing — Edit images via OpenAI, xAI, Gemini, or BFL (FLUX Kontext)
- 🎬 Video Generation — Generate videos via OpenAI (sora-2), xAI (grok-imagine-video), or Gemini (veo-3.1)
- 🔊 Audio Generation — Text-to-speech via OpenAI (tts-1), Gemini, or ElevenLabs (Flash v2.5). Sound effects via ElevenLabs
- 🎙️ Audio Transcription — Speech-to-text via OpenAI (Whisper) or ElevenLabs (Scribe)
- 🔄 Auto-Discovery — Automatically detects configured providers from environment variables
- 🎯 Provider Selection — Auto-selects or explicitly choose a provider per request
- 📁 File Output — Saves all generated media to disk with descriptive filenames
Quick Start
Set the API key for at least one provider. Most users only need one — add more to access additional providers.
# Using OpenAI
claude mcp add multimodal-mcp -e OPENAI_API_KEY=sk-... -- npx -y @r16t/multimodal-mcp@latest
# Or using xAI
# claude mcp add multimodal-mcp -e XAI_API_KEY=xai-... -- npx -y @r16t/multimodal-mcp@latest
# Or using Gemini
# claude mcp add multimodal-mcp -e GEMINI_API_KEY=AIza... -- npx -y @r16t/multimodal-mcp@latest
# Or using ElevenLabs (audio + transcription)
# claude mcp add multimodal-mcp -e ELEVENLABS_API_KEY=xi-... -- npx -y @r16t/multimodal-mcp@latest
# Or using BFL/FLUX (images)
# claude mcp add multimodal-mcp -e BFL_API_KEY=... -- npx -y @r16t/multimodal-mcp@latest
Using a different editor? See setup instructions for Claude Desktop, Cursor, VS Code, Windsurf, and Cline.
Environment Variables
| Variable | Required | Description |
|---|---|---|
OPENAI_API_KEY |
At least one provider key | OpenAI API key — enables image, video, audio generation, and transcription via gpt-image-1, sora-2, tts-1, and whisper-1 |
XAI_API_KEY |
At least one provider key | xAI API key — enables image and video generation via grok-imagine-image and grok-imagine-video |
GEMINI_API_KEY |
At least one provider key | Gemini API key — enables image, video, and audio generation via imagen-4, veo-3.1, and gemini-2.5-flash-preview-tts |
GOOGLE_API_KEY |
— | Alias for GEMINI_API_KEY; either name is accepted |
ELEVENLABS_API_KEY |
At least one provider key | ElevenLabs API key — enables audio generation (TTS, sound effects) and transcription via Flash v2.5 and Scribe v1 |
BFL_API_KEY |
At least one provider key | BFL API key — enables image generation and editing via FLUX Pro 1.1 and FLUX Kontext |
MEDIA_OUTPUT_DIR |
No | Directory for saved media files. Defaults to the current working directory |
Available Tools
generate_image
Generate an image from a text prompt.
| Parameter | Type | Required | Description |
|---|---|---|---|
prompt |
string | Yes | Text description of the image to generate |
provider |
string | No | Provider to use: openai, xai, google, bfl. Auto-selects if omitted |
aspectRatio |
string | No | Aspect ratio: 1:1, 16:9, 9:16, 4:3, 3:4 |
quality |
string | No | Quality level: low, standard, high |
outputDirectory |
string | No | Directory to save the generated file. Absolute or relative path. Defaults to MEDIA_OUTPUT_DIR or cwd |
providerOptions |
object | No | Provider-specific parameters passed through directly |
generate_video
Generate a video from a text prompt. Video generation is asynchronous and may take several minutes.
| Parameter | Type | Required | Description |
|---|---|---|---|
prompt |
string | Yes | Text description of the video to generate |
provider |
string | No | Provider to use: openai, xai, google. Auto-selects if omitted |
duration |
number | No | Video duration in seconds (provider limits apply) |
aspectRatio |
string | No | Aspect ratio: 16:9, 9:16, 1:1 |
resolution |
string | No | Resolution: 480p, 720p, 1080p |
outputDirectory |
string | No | Directory to save the generated file. Absolute or relative path. Defaults to MEDIA_OUTPUT_DIR or cwd |
providerOptions |
object | No | Provider-specific parameters passed through directly |
generate_audio
Generate audio from text. Supports text-to-speech and sound effects. Audio generation is synchronous.
| Parameter | Type | Required | Description |
|---|---|---|---|
text |
string | Yes | Text to convert to speech, or a description of the sound effect to generate |
provider |
string | No | Provider to use: openai, google, elevenlabs. Auto-selects if omitted |
voice |
string | No | Voice name (provider-specific). OpenAI: alloy, ash, coral, echo, fable, nova, onyx, sage, shimmer. Google: Kore, Charon, Fenrir, Aoede, Puck, etc. ElevenLabs: voice ID |
speed |
number | No | Speech speed multiplier (OpenAI only): 0.25 to 4.0 |
format |
string | No | Output format (OpenAI only): mp3, opus, aac, flac, wav, pcm |
outputDirectory |
string | No | Directory to save the generated file. Absolute or relative path. Defaults to MEDIA_OUTPUT_DIR or cwd |
providerOptions |
object | No | Provider-specific parameters passed through directly. ElevenLabs: set mode: "sound-effect" for sound effects, model for TTS model selection |
transcribe_audio
Transcribe audio to text (speech-to-text).
| Parameter | Type | Required | Description |
|---|---|---|---|
audioPath |
string | Yes | Absolute path to the audio file to transcribe |
provider |
string | No | Provider to use: openai, elevenlabs. Auto-selects if omitted |
language |
string | No | Language code (e.g., en, fr, es) to hint the transcription language |
providerOptions |
object | No | Provider-specific parameters passed through directly |
list_providers
List all configured media generation providers and their capabilities. Takes no parameters.
Provider Capabilities
| Provider | Image | Image Editing | Video | Audio | Transcription | Key Models |
|---|---|---|---|---|---|---|
| OpenAI | ✅ | ✅ | ✅ | ✅ | ✅ | gpt-image-1, sora-2, tts-1, whisper-1 |
| xAI | ✅ | ✅ | ✅ | — | — | grok-imagine-image, grok-imagine-video |
| Gemini | ✅ | ✅ | ✅ | ✅ | — | imagen-4, veo-3.1, gemini-2.5-flash-preview-tts |
| ElevenLabs | — | — | — | ✅ | ✅ | eleven_flash_v2_5, scribe_v1 |
| BFL | ✅ | ✅ | — | — | — | flux-pro-1.1, flux-kontext-pro |
Image Aspect Ratios
| Provider | 1:1 | 16:9 | 9:16 | 4:3 | 3:4 |
|---|---|---|---|---|---|
| OpenAI | ✅ | ✅ | ✅ | ✅ | ✅ |
| xAI | ✅ | ✅ | ✅ | ✅ | ✅ |
| Gemini | ✅ | ✅ | ✅ | ✅ | ✅ |
| BFL | ✅ | ✅ | ✅ | ✅ | ✅ |
Video Aspect Ratios & Resolutions
| Provider | 16:9 | 9:16 | 1:1 | 480p | 720p | 1080p |
|---|---|---|---|---|---|---|
| OpenAI | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ |
| xAI | ✅ | ✅ | ✅ | — | ✅ | ✅ |
| Gemini | ✅ | ✅ | — | — | ✅ | ✅ |
Audio Formats
| Provider | mp3 | opus | aac | flac | wav | pcm |
|---|---|---|---|---|---|---|
| OpenAI | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ |
| Gemini | — | — | — | — | ✅ | — |
| ElevenLabs | ✅ | ✅ | — | — | — | ✅ |
Troubleshooting
No providers configured
[config] No provider API keys detected
Set at least one of OPENAI_API_KEY, XAI_API_KEY, GEMINI_API_KEY, ELEVENLABS_API_KEY, or BFL_API_KEY in the MCP server's env block.
Provider not available for requested media type
Each provider supports different media types (see Provider Capabilities). If you specify a provider that isn't configured (no API key) or doesn't support the requested media type, you'll receive an error. Omit the provider parameter to auto-select from configured providers.
Video generation timeout
Video generation polls for up to 10 minutes. If your video hasn't completed in that window, the request will fail with a timeout error. Try a shorter duration or a simpler prompt.
xAI image generation returned no data
This indicates the xAI API returned an empty response. Check that your XAI_API_KEY is valid and that your prompt does not violate xAI content policies.
Gemini image/video generation failed: 403
Verify your GEMINI_API_KEY has the Generative Language API enabled in Google Cloud Console.
Development
npm run build # Compile TypeScript to build/
npm test # Run tests with Vitest
npm run lint # Lint and auto-fix with ESLint
npm run typecheck # Type-check without emitting
npm run dev # Watch mode for TypeScript compilation
Editor Setup
Replace OPENAI_API_KEY with your provider of choice (XAI_API_KEY, GEMINI_API_KEY, ELEVENLABS_API_KEY, BFL_API_KEY). You can set multiple keys to enable multiple providers.
Claude Desktop
Add to ~/Library/Application Support/Claude/claude_desktop_config.json:
{
"mcpServers": {
"multimodal-mcp": {
"command": "npx",
"args": ["@r16t/multimodal-mcp@latest"],
"env": {
"OPENAI_API_KEY": "sk-..."
}
}
}
}
Cursor
Add to .cursor/mcp.json in your project root (or ~/.cursor/mcp.json globally):
{
"mcpServers": {
"multimodal-mcp": {
"command": "npx",
"args": ["@r16t/multimodal-mcp@latest"],
"env": {
"OPENAI_API_KEY": "sk-..."
}
}
}
}
VS Code (GitHub Copilot)
Add to .vscode/mcp.json in your project root:
{
"servers": {
"multimodal-mcp": {
"command": "npx",
"args": ["@r16t/multimodal-mcp@latest"],
"env": {
"OPENAI_API_KEY": "sk-..."
}
}
}
}
Windsurf
Add to ~/.codeium/windsurf/mcp_config.json:
{
"mcpServers": {
"multimodal-mcp": {
"command": "npx",
"args": ["@r16t/multimodal-mcp@latest"],
"env": {
"OPENAI_API_KEY": "sk-..."
}
}
}
}
Cline
Add to ~/Library/Application Support/Code/User/globalStorage/saoudrizwan.claude-dev/settings/cline_mcp_settings.json:
{
"mcpServers": {
"multimodal-mcp": {
"command": "npx",
"args": ["@r16t/multimodal-mcp@latest"],
"env": {
"OPENAI_API_KEY": "sk-..."
}
}
}
}
License
MIT
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。