Scenario

automate

MCP

Your entire Scenario workspace, accessible from your AI agent. Generate, train, manage, and ship - without switching tools.

Try it nowHelp guide

Key benefits

  • Full creative platform, from your agent

    Every capability available in the Scenario web app is accessible from your AI agent, including image, video, 3D, and audio generation, custom model training, and composing LoRAs - no manual API calls required.

  • 500+ models from 50+ providers

    Access the full model library through a single MCP server, covering providers including Black Forest Labs, Google, OpenAI, Alibaba, Bytedance, Runway ML, ElevenLabs, Tencent, Meta, MiniMax, Luma Labs, and more.

  • Full pipeline in natural language

    From generation to asset management and workflow execution, everything runs from a single agent conversation. Describe what you need, and the MCP server handles model selection, generation, and delivery.

How it works

  1. 1

    Install the MCP server

    Add Scenario to your agent in one command. Supported in Claude Code, Cursor, VS Code, Claude Desktop, Windsurf, Zed, Warp, Gemini CLI, and more.

  2. 2

    Authenticate your account

    Connect via OAuth for a one-click browser flow (recommended), or use an API key for static credentials. OAuth requires no config - just make your first request and a browser tab opens automatically.

  3. 3

    Generate in natural language

    Ask your agent anything: generate an image, train a model, search your assets, or trigger a workflow. The MCP server picks the right tool and returns the result inline.

Use cases

  • Automated game asset production

    Generate on-brand characters, props, environments, and UI elements directly from your build pipeline, without switching to the Scenario web app or running manual generation sessions.

  • Dynamic content generation in applications

    Trigger real-time image or video generation from within your IDE while building web apps, mobile products, or internal tools - keeping your creative and coding workflows in one place.

  • Train and deploy custom models from your agent

    Upload reference images, start a training run, and generate with your custom LoRA model entirely through natural language. No context switching between tools.