MeiGen AI Design MCP

What jau123/MeiGen-AI-Design-MCP does

jau123/MeiGen-AI-Design-MCP exposes multimedia creation workflows through MCP. Its tools cover AI image generation, MeiGen video generation, prompt enhancement, semantic gallery search, model discovery, image uploads, background removal, product-detail imagery, marketing posters, AI-generated product backgrounds, and image upscaling.

The server supports three image-generation routes: MeiGen, local ComfyUI, and OpenAI-compatible APIs. MeiGen Cloud is used for video generation and the specialized visual workflows. A prompt can be developed from a short idea, refined with gallery inspiration, or combined with reference images supplied as local files or public HTTPS URLs.

How it works

Use search_gallery to find visually or conceptually related prompts and reference images, or call enhance_prompt to expand a short description into a more detailed image prompt. list_models reports available image-generation models and capabilities. list_skills exposes workflow requirements, upload options, current prices, and account setup information for specialized operations.

Generation and paid visual skills return structured status information, handles, errors, and completed resource URLs. Long-running jobs can be inspected with check_generation or check_skill; callers should follow the returned next action and wait interval rather than submit duplicate work. The caller is responsible for retaining request IDs, handling progress, and presenting or downloading results.

Specialized tools accept actual local files or accessible public URLs. Product-detail generation creates one to six coordinated ecommerce images, while poster generation needs only a subject and can optionally use supplied copy, dates, logos, product photos, or style references. Background replacement accepts a product photo and supports white, smart, or custom backgrounds.

Setup and configuration

The repository supports a local npm stdio configuration using Node.js 22 or newer as the recommended runtime:

{
  "mcpServers": {
    "meigen": {
      "command": "npx",
      "args": ["-y", "meigen@3.3.3"]
    }
  }
}

A local API-key setup uses MEIGEN_API_TOKEN in the server’s private environment. MeiGen Cloud generation and the visual Skills require purchased credits. Prompt enhancement and gallery search do not require an API key. The remote deployment is available over Streamable HTTP at https://www.meigen.ai/api/mcp, where clients authenticate with MeiGen through OAuth.

Tools and capabilities

• Generate images through MeiGen, ComfyUI, or OpenAI-compatible image providers.
• Generate MeiGen videos using a live model ID from list_models.
• Search a semantic gallery and retrieve inspiration prompts and image URLs.
• Remove backgrounds, create product scenes, upscale still images, and build marketing or ecommerce visuals.
• Manage ComfyUI workflow templates and user preferences.
• Upload reference images and inspect prices, models, skills, job status, and recovery states.

Limitations and notes

MeiGen Cloud generation consumes purchased credits; estimates are not enforced spending caps, and batches are not atomic. Image inputs are subject to a 64 MiB and 64 MP source limit. Inputs above 4096 pixels or 16 MP may be resized when accepted. Upscaling applies to still images, not video.

Private MeiGen libraries, favorites, boards, history, and saved references require remote OAuth or the official ChatGPT integration; local API-key connections cannot access them. Creating saved characters is not supported. Temporary uploaded media URLs are cleared after seven days. Video submissions are limited to four concurrent jobs per MCP process, subject to backend quotas and retry timing.