openapi-integrator-mcp

OpenAI Compatible API MCP Server (English)

This project is a Model Context Protocol (MCP) server that integrates various tools based on OpenAI compatible APIs and SiliconFlow APIs. It offers a range of functionalities including image generation, image editing, speech synthesis (TTS), speech-to-text (STT), and video generation.

Core Features

• Image Generation: Generate images using OpenAI compatible APIs (e.g., DALL-E 3, gpt-image-1, or other Stable Diffusion models).
• Image Editing: Edit images using OpenAI compatible APIs (e.g., gpt-image-1).
• Speech Synthesis (TTS): Convert text into speech using OpenAI compatible APIs.
• Speech-to-Text (STT): Transcribe audio files into text using OpenAI compatible APIs.
• Video Generation: Submit text-to-video or image-to-video generation tasks using the SiliconFlow API.
• Background Task Processing: For time-consuming tasks (like image generation/editing with specific models, video generation), the server accepts the task and processes it asynchronously. Notifications are sent via configured channels (OneBot or Telegram) upon completion or failure.
• File Upload: Supports uploading generated images and videos to a configured ImgBed service based on MarSeventh/CloudFlare-ImgBed.
• Local Storage: All generated media files are saved locally.

Prerequisites

• Node.js (recommended version >= 18.x)
• Relevant API keys (OpenAI/compatible API key, SiliconFlow API key, etc.)

Installation and Setup

1. Clone the project (if you haven't already):

   git clone <project_repository_url>
   cd openapi-integrator-mcp
   

2. Install dependencies:

   npm install
   

3. Build the project:

   npm run build
   

   The build output will be in the build directory.

Configuration

Configure environment variables via a .env file in the project root. If the .env file does not exist, create it based on .env.example (if provided) or the list below.

Key Environment Variables:

• OPENAI_API_KEY: (Required) Your OpenAI API key or a compatible API key.
• OPENAI_API_BASE_URL: (Optional) Base URL for the OpenAI compatible API. Defaults to https://api.openai.com.
• REQUEST_TIMEOUT: (Optional) API request timeout in milliseconds. Defaults to 180000 (3 minutes).
• OUTPUT_DIR: (Optional) Base output directory for generated media files. Defaults to ./output. Subdirectories like images, audio, video will be created here.

Default Model Configuration for Image Generation/Editing:

• DEFAULT_IMAGE_MODEL: (Optional) Default image generation model. Defaults to dall-e-3.
• DEFAULT_EDIT_IMAGE_MODEL: (Optional) Default image editing model. Defaults to gpt-image-1.
• DEFAULT_IMAGE_WIDTH: (Optional) Default image width (for non-DALL-E 3/gpt-image-1 models). Defaults to 1024.
• DEFAULT_IMAGE_HEIGHT: (Optional) Default image height (for non-DALL-E 3/gpt-image-1 models). Defaults to 768.
• DEFAULT_IMAGE_STEPS: (Optional) Default image generation steps (for non-DALL-E 3/gpt-image-1 models). Defaults to 20.

Default Speech Configuration:

• DEFAULT_SPEECH_MODEL: (Optional) Default speech synthesis model. Defaults to tts-1.
• DEFAULT_SPEECH_VOICE: (Optional) Default speech synthesis voice. Defaults to alloy.
• DEFAULT_SPEECH_SPEED: (Optional) Default speech synthesis speed. Defaults to 1.0.
• DEFAULT_TRANSCRIPTION_MODEL: (Optional) Default speech transcription model. Defaults to whisper-1.

SiliconFlow Video Generation Configuration:

• SILICONFLOW_API_KEY: (Optional, Required if using video generation) SiliconFlow API key.
• SILICONFLOW_BASE_URL: (Optional) SiliconFlow API base URL. Defaults to https://api.siliconflow.cn.
• SILICONFLOW_VIDEO_MODEL: (Optional) Default video generation model. Defaults to Wan-AI/Wan2.1-T2V-14B-Turbo.

Notification Configuration (configure at least one to receive background task results):

• ONEBOT_HTTP_URL: (Optional) OneBot HTTP post URL (e.g., http://localhost:5700).
• ONEBOT_ACCESS_TOKEN: (Optional) OneBot Access Token (if required).
• ONEBOT_MESSAGE_TYPE: (Optional) OneBot message type (private or group).
• ONEBOT_TARGET_ID: (Optional) OneBot target user ID or group ID.
• TELEGRAM_BOT_TOKEN: (Optional) Telegram Bot Token.
• TELEGRAM_CHAT_ID: (Optional) Telegram Chat ID.

ImgBed Configuration (Optional, for MarSeventh/CloudFlare-ImgBed):

• CF_IMGBED_UPLOAD_URL: (Optional) Your deployed CloudFlare-ImgBed upload URL (e.g., https://your-worker.your-domain.workers.dev/upload).
• CF_IMGBED_API_KEY: (Optional) The AUTH_KEY (or authCode as referred to in some contexts) configured for your CloudFlare-ImgBed instance.

Running the Server

• Production Mode:

  npm start
  

  This will run the JavaScript files from the build directory.

• Development Mode (uses ts-node-dev for hot-reloading):

  npm run dev
  

Once started, the MCP server will listen for requests on standard input/output (stdio).

Available MCP Tools

1. generate_image

Generates an image.

• Function: Creates an image based on a text prompt using OpenAI compatible APIs. Supports various models like DALL-E 3, gpt-image-1, and others. For DALL-E 3/gpt-image-1, tasks are processed in the background with results sent via notification. Other models return results synchronously.
• Key Parameters:
  • prompt (string, required): Description of the image.
  • model (string, optional): Model to use, defaults to DEFAULT_IMAGE_MODEL from config.
  • n (number, optional): Number of images to generate.
  • (DALL-E 3/gpt-image-1 specific): quality, size, background, moderation.
  • (Other models specific): width, height, steps.

2. edit_image

Edits an image.

• Function: Modifies an existing image based on a text prompt using models like gpt-image-1. Tasks are processed in the background with results sent via notification.
• Key Parameters:
  • image (string, required): Path or URL of the image to edit.
  • prompt (string, required): Editing instructions.
  • model (string, optional): Model to use, defaults to DEFAULT_EDIT_IMAGE_MODEL from config.
  • n (number, optional): Number of images to generate.
  • size (string, optional): Output image size.

3. generate_speech

Text-to-Speech.

• Function: Converts text into an audio file (MP3 format) and saves it locally.
• Key Parameters:
  • input (string, required): Text to convert to speech.
  • voice (string, required): Voice to use.
  • model (string, optional): Model to use, defaults to DEFAULT_SPEECH_MODEL from config.
  • speed (number, optional): Speech speed, defaults to DEFAULT_SPEECH_SPEED from config.

4. transcribe_audio

Speech-to-Text.

• Function: Transcribes an audio file into text.
• Key Parameters:
  • file (string, required): Local path or URL of the audio file.
  • model (string, optional): Model to use, defaults to DEFAULT_TRANSCRIPTION_MODEL from config.

5. generate_video

Generates a video.

• Function: Submits a video generation task to the SiliconFlow API. Supports text-to-video and image-to-video. Tasks are processed in the background with results sent via notification.
• Key Parameters:
  • prompt (string, required): Description of the video.
  • image_size (string, required): Video dimensions/aspect ratio (e.g., "1280x720").
  • model (string, optional): Video model to use, defaults to SILICONFLOW_VIDEO_MODEL from config. Supported models include: Wan-AI/Wan2.1-T2V-14B, Wan-AI/Wan2.1-T2V-14B-Turbo, Wan-AI/Wan2.1-I2V-14B-720P, Wan-AI/Wan2.1-I2V-14B-720P-Turbo.
  • image (string, optional): Image URL or Base64 encoded data, required for Image-to-Video models.
  • negative_prompt (string, optional): Negative prompt.
  • seed (integer, optional): Random seed.

Notification Feature

For tasks processed in the background (e.g., specific image generation/editing, video generation), results will be sent via notifications configured through:

• OneBot: If ONEBOT_HTTP_URL, ONEBOT_MESSAGE_TYPE, and ONEBOT_TARGET_ID are configured.
• Telegram: If TELEGRAM_BOT_TOKEN and TELEGRAM_CHAT_ID are configured.

Please ensure at least one notification method is configured to receive results for background tasks.

License

This project is licensed under the MIT License. See the LICENSE file for details.