Krea Server

npm version License: Apache 2.0 MCP CI GitHub stars

MCP server for Krea.ai - Generate stunning AI images and videos using Flux, Hailuo, Runway, Kling, Ideogram, Imagen, and 20+ other state-of-the-art models.

212+ developers have cloned this repo within 2 weeks. Works with Claude Desktop, Claude Code, Cursor, and any MCP-compatible client.

---

Quick Start (No installation required!)

You can run the MCP server directly via npx:

{
  "mcpServers": {
    "krea": {
      "command": "npx",
      "args": [
        "-y",
        "@vmosaic/krea-mcp-server"
      ],
      "env": {
        "KREA_API_KEY": "YOUR_KREA_API_KEY"
      }
    }
  }
}

Why Krea MCP?

• 20+ AI Models - Access Flux, Ideogram, Imagen 4, Runway Gen-4, Hailuo, Kling, Pika, Veo 3, and more through one unified interface
• Text-to-Image - Generate photorealistic images, art, illustrations from text prompts
• Text-to-Video - Create AI videos with natural motion and cinematic quality
• Image-to-Video - Animate any image with AI-powered motion
• Custom Styles - Apply LoRA styles for consistent aesthetics
• Simple Setup - One command install, works instantly with Claude

---

Optional Global Install

1. Install

npm install -g @vmosaic/krea-mcp-server

2. Get API Key

Get your API key from krea.ai

3. Run the global binary

krea-mcp

4. Configure Claude Desktop with the global binary

Add to ~/.config/claude/claude_desktop_config.json (Linux) or ~/Library/Application Support/Claude/claude_desktop_config.json (macOS):

{
  "mcpServers": {
    "krea": {
      "command": "krea-mcp",
      "env": {
        "KREA_API_KEY": "your-api-key-here"
      }
    }
  }
}

4b. Or configure Claude Code without a global install

claude mcp add krea -e KREA_API_KEY=your-api-key -- npx -y @vmosaic/krea-mcp-server

---

Features

Image Generation

Generate images with state-of-the-art AI models:

"Generate a cyberpunk cityscape at sunset with neon lights reflecting on wet streets"

Supported models: Flux Dev, Flux Pro, Flux Schnell, Ideogram V2, Imagen 4, Krea 1, ChatGPT Image, Topaz, Bloom, and more.

Video Generation

Create AI videos from text or images:

"Create a video of a golden retriever running through a field of sunflowers"

Supported models: Hailuo, Kling 1.6, Runway Gen-4, Pika 2, Veo 3, Sora 2, Seedance, Ray 2, and more.

Style Transfer

Apply custom styles (LoRAs) to your generations:

"Generate an image in anime style of a samurai in cherry blossom garden"

Asset Management

Upload and manage your images for image-to-video and other workflows.

---

Available Tools

Tool	Description
generate_image	Generate images from text prompts
generate_video	Generate videos from text or images
get_job	Check generation job status
list_jobs	List recent generation jobs
upload_asset	Upload images/videos for use in generations
get_asset	Get asset details
list_assets	List uploaded assets
search_styles	Search for styles/LoRAs
get_style	Get style details

---

Claude Code Skill

This repo includes a /krea slash command for Claude Code.

Install the skill:

cp -r .claude/skills/krea ~/.claude/skills/

Usage:

/krea a photorealistic portrait of a astronaut on mars
/krea video ocean waves crashing on rocks at golden hour
/krea status job_abc123
/krea styles cyberpunk

---

Supported Models

Image Models

Model	ID	Best For
FLUX.1 Dev	flux-dev	High quality, balanced
FLUX.1 Pro	flux-pro	Highest quality
FLUX.1 Schnell	flux-schnell	Fast generation
Ideogram V2	ideogram-v2	Text in images
Imagen 4	imagen-4	Photorealism
Krea 1	krea-1	Creative styles
ChatGPT Image	chatgpt-image	General purpose

Video Models

Model	ID	Best For
Hailuo	hailuo	High quality, natural motion
Kling 1.6	kling-1.6	Detailed motion
Runway Gen-4	runway-gen4	Cinematic quality
Pika 2	pika-2	Creative styles
Veo 3	veo-3	Google's latest
Sora 2	sora-2	OpenAI's model

---

API Reference

generate_image

{
  prompt: string,        // Required: Image description
  model?: string,        // Default: "flux-dev"
  width?: number,        // Default: 1024
  height?: number,       // Default: 1024
  style_id?: string,     // Optional: Style/LoRA ID
  negative_prompt?: string
}

generate_video

{
  prompt: string,        // Required: Video description
  model?: string,        // Default: "hailuo"
  image_url?: string,    // Optional: First frame for i2v
  duration?: number,     // Default: 5 seconds
  aspect_ratio?: string  // Default: "16:9"
}

---

Examples

Generate a Product Photo

Use Krea to generate a professional product photo of a perfume bottle
on a marble surface with soft studio lighting

Create a Social Media Video

Generate a 5-second video of coffee being poured into a cup in slow motion,
cinematic lighting, 9:16 aspect ratio for Instagram Reels

Apply a Custom Style

Search for "watercolor" styles and generate an image of a Paris street scene
using that style

---

Links

• Krea.ai - Official Krea website
• Krea API Docs - API documentation
• MCP Specification - Model Context Protocol
• Claude Desktop - Download Claude Desktop

---

Contributing

Contributions welcome! Please feel free to submit a Pull Request.

---

License

Apache License 2.0 - see LICENSE for details.

---

Keywords

mcp, mcp-server, model-context-protocol, krea, krea-ai, ai, artificial-intelligence, image-generation, video-generation, text-to-image, text-to-video, image-to-video, flux, ideogram, imagen, runway, hailuo, kling, pika, veo, sora, generative-ai, ai-art, ai-video, claude, anthropic, claude-desktop, claude-code, cursor, llm, llm-tools, ai-tools, machine-learning, deep-learning, stable-diffusion, diffusion-models