MCP Server

MCP (Model Context Protocol) lets AI assistants use RendShot tools directly. Generate images and capture screenshots from natural language prompts.

Client Setup

Claude Desktop

Add the following to your Claude Desktop MCP configuration file:

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
• Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "rendshot": {
      "command": "npx",
      "args": ["-y", "@rendshot/mcp"],
      "env": {
        "RENDSHOT_API_KEY": "rs_live_YOUR_KEY"
      }
    }
  }
}

Restart Claude Desktop after saving the configuration.

Claude Code

Register the MCP server with your API key:

claude mcp add rendshot -e RENDSHOT_API_KEY=rs_live_YOUR_KEY -- npx -y @rendshot/mcp

Cursor

Add to your project-level .cursor/mcp.json or global ~/.cursor/mcp.json:

{
  "mcpServers": {
    "rendshot": {
      "command": "npx",
      "args": ["-y", "@rendshot/mcp"],
      "env": {
        "RENDSHOT_API_KEY": "rs_live_YOUR_KEY"
      }
    }
  }
}

Windsurf

Edit ~/.codeium/windsurf/mcp_config.json or open Windsurf Settings > MCP:

{
  "mcpServers": {
    "rendshot": {
      "command": "npx",
      "args": ["-y", "@rendshot/mcp"],
      "env": {
        "RENDSHOT_API_KEY": "rs_live_YOUR_KEY"
      }
    }
  }
}

VS Code

Create a .vscode/mcp.json file in your project root:

{
  "servers": {
    "rendshot": {
      "type": "stdio",
      "command": "npx",
      "args": ["-y", "@rendshot/mcp"],
      "env": {
        "RENDSHOT_API_KEY": "rs_live_YOUR_KEY"
      }
    }
  }
}

Or install via command line:

code --add-mcp '{"name":"rendshot","command":"npx","args":["-y","@rendshot/mcp"],"env":{"RENDSHOT_API_KEY":"rs_live_YOUR_KEY"}}'

Cline

Open Cline Settings > MCP Servers > Edit Config in VS Code and add:

{
  "mcpServers": {
    "rendshot": {
      "command": "npx",
      "args": ["-y", "@rendshot/mcp"],
      "env": {
        "RENDSHOT_API_KEY": "rs_live_YOUR_KEY"
      }
    }
  }
}

Environment Variables

Available Tools

generate_image

Render HTML/CSS, a template, or an AI-generated design from a natural language prompt. Provide one of html, template_id, or prompt. prompt + template_id is allowed as a visual style reference. prompt + html is not allowed.

* Provide exactly one of html, template_id, or prompt. prompt + template_id is also allowed (style reference).

When called with prompt, the tool response's structuredContent includes the AI-generated html and variables alongside the rendered image — feed those straight into create_template to save the design for reuse. See the AI Render guide for the full loop.

screenshot_url

Capture a screenshot of a public URL. Defaults to 1280 x 800.

list_templates

Browse available image templates. Returns template IDs, names, dimensions, and variable definitions.

get_template

Get detailed information about a template, including its variable definitions. Use this before generate_image to know which variables to pass.

create_template

Save an HTML template for reuse. Typically used after generate_image returns an AI-generated template you want to keep. Templates are created as private drafts — publish them through the web dashboard if you want them in the public gallery.

Recommended workflow:

1. Call generate_image with a prompt to get an AI-generated design.
2. If the assistant is happy with the result, call create_template with the returned html and variables.
3. Use the new template_id in future generate_image calls for faster, deterministic, cheaper renders.

Example Prompts

Once the MCP server is configured, ask your AI assistant directly:

• "Screenshot https://example.com at 1440 wide"
• "Render this HTML as a social card image"
• "Take a full-page screenshot of my staging site"
• "List available templates for Twitter"
• "Use the Social Card template with title 'Hello World'"
• "Generate a Xiaohongshu cover for a new coffee product, then save it as a template called 'Coffee Cover'"
• "Make me an OG image announcing the v2 launch, minimal dark style"

Next Steps

• JavaScript SDK -- Full SDK reference with all options
• Python SDK -- Official Python client
• CLI -- Use RendShot from the command line
• API Reference -- Full endpoint documentation