MCP Directory

How to add MeiGen AI Design MCP to Windsurf

MCP server for AI image & video generation across 9 models, with a 1,446-prompt library and parallel sub-agent orchestration. Paste the config into ~/.codeium/windsurf/mcp_config.json and restart Windsurf.

Last updated June 14, 2026 · 1.5k · stdio · apikey

Windsurf config for MeiGen AI Design MCP

npx -y meigen@1.3.2
{
  "mcpServers": {
    "meigen-ai-design-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "meigen@1.3.2"
      ],
      "env": {
        "MEIGEN_API_TOKEN": "meigen_sk_..."
      }
    }
  }
}

Setup steps

  1. 1Open Windsurf → Cascade → the hammer/MCP icon → Configure (or edit ~/.codeium/windsurf/mcp_config.json).
  2. 2Paste the MeiGen AI Design MCP config below.
  3. 3Fill in placeholder secrets, then save.
  4. 4Click Refresh in the MCP panel.
  5. 5MeiGen AI Design MCP's tools become available to Cascade.

Before you start

  • Node.js with npx (server runs via `npx -y meigen@1.3.2`)
  • An MCP-compatible host (Claude Code, Cursor, VS Code, Windsurf, Roo Code, OpenClaw, Hermes Agent, etc.) — or use the standalone CLI
  • For image/video generation: a MeiGen API token (meigen_sk_...), an OpenAI-compatible API key + base URL + model, or a local ComfyUI server (free features need no key)

What MeiGen AI Design MCP can do in Windsurf

search_gallery

Search 1,446 curated trending prompts with visual previews (powered by nanobanana-trending-prompts). Free.

get_inspiration

Get the full prompt, all images, and metadata for any gallery entry. Free.

enhance_prompt

Transform a brief idea into a professional image prompt. Free.

list_models

List available models across all configured providers. Free.

comfyui_workflow

Manage ComfyUI workflow templates: list, view, import, modify, delete. Free.

manage_preferences

Remember your preferred style, aspect ratio, model, and favorite prompts. Free.

generate_image

Generate an image — routes to the best available provider automatically. Local reference images are auto-compressed and uploaded. Requires a provider key.

generate_video

Generate a video via the MeiGen platform — Seedance 2.0 (fast/pro), Happyhorse 1.0, or Veo 3.1. Supports text-to-video and first-frame image-to-video; local files auto-uploaded. Saves MP4 to ~/Movies/meigen/. Requires a provider key.

Security

Free features (gallery search, prompt enhancement, model listing) run locally against bundled data with no external calls. Image/video generation requires a provider key (MEIGEN_API_TOKEN, OPENAI_API_KEY, or a ComfyUI URL). With MeiGen Cloud or an OpenAI-compatible endpoint, prompts and reference images are sent to that API; ComfyUI keeps all processing on your machine. Reference images are compressed locally (max 2MB) and uploaded to Cloudflare R2 via gen.meigen.ai (no auth, expire after 24h); ComfyUI users can skip upload by passing local paths. A custom S3/R2 upload backend can be configured via UPLOAD_GATEWAY_URL. The project states no telemetry, analytics, or tracking.

MeiGen AI Design MCP + Windsurf FAQ

Where is the Windsurf config file?

Windsurf reads MCP servers from ~/.codeium/windsurf/mcp_config.json. Paste the MeiGen AI Design MCP config there under the "mcpServers" key and restart the client.

Is MeiGen AI Design MCP safe to use with Windsurf?

Free features (gallery search, prompt enhancement, model listing) run locally against bundled data with no external calls. Image/video generation requires a provider key (MEIGEN_API_TOKEN, OPENAI_API_KEY, or a ComfyUI URL). With MeiGen Cloud or an OpenAI-compatible endpoint, prompts and reference images are sent to that API; ComfyUI keeps all processing on your machine. Reference images are compressed locally (max 2MB) and uploaded to Cloudflare R2 via gen.meigen.ai (no auth, expire after 24h); ComfyUI users can skip upload by passing local paths. A custom S3/R2 upload backend can be configured via UPLOAD_GATEWAY_URL. The project states no telemetry, analytics, or tracking.

Do I need an API key to use it?

No, not for everything. Free features — gallery search, prompt enhancement, model listing, ComfyUI workflow management, and preference memory — work immediately with no key. Only generate_image and generate_video require a provider key (MeiGen token, an OpenAI-compatible key, or a local ComfyUI URL).

Which backends are supported?

Three: MeiGen Cloud (9 image/video models, no GPU needed), any OpenAI-compatible API (bring your own key, base URL, and model — Together AI, Fireworks AI, DeepInfra, SiliconFlow, etc.), and local ComfyUI (offline, runs on your GPU). You can configure one or several; the server auto-selects the best available.

Which hosts and tools does it work with?

Claude Code (via plugin marketplace or .mcp.json), Cursor, VS Code / GitHub Copilot, Windsurf, Roo Code (one-command setup via `npx meigen init <tool>`), OpenClaw (ClawHub bundle), Hermes Agent, and any MCP-compatible host. It also ships a standalone CLI that needs no MCP host.

View repo Full MeiGen AI Design MCP page