imagekit-mcp-server 1.0.0

ImageKit Public MCP Server - exposes search_docs and transformation_builder tools

ImageKit Public MCP Server

A Python MCP (Model Context Protocol) server that exposes ImageKit's documentation search and transformation builder as tools. Deployable to AWS Lambda via SAM.

Tools

search_docs

Search ImageKit documentation across guides, API references, SDK docs, and community content.

Parameters:

• query (required): Search query string
• sources (optional): Filter by source — imagekit_api_references, imagekit_community, imagekit_guides, imagekit_sdk

transformation_builder

Build ImageKit image/video transformation URLs from natural language descriptions.

Parameters:

• query (required): Natural language description (e.g., "resize to 300x200 and add blur")
• src (optional): Source ImageKit URL to apply transformations to
• fetch_url_to_check (optional, default: true): Verify the generated URL works

Prerequisites

• Python 3.12+
• uv (Python package manager)
• AWS CLI + SAM CLI (for deployment)
• Docker (for SAM build)

Local Development

# Install dependencies
uv sync

# Copy and configure environment
cp .env.example .env

# Run the server locally (port 8000)
uv run python -m src.server

# Run tests
uv run pytest

# Lint
uv run ruff check src/ tests/

The MCP server will be available at:

• SSE endpoint: http://localhost:8000/sse
• Messages endpoint: http://localhost:8000/messages/
• Health check: http://localhost:8000/health

Deployment (AWS SAM)

# First time (guided setup)
./scripts/deploy.sh --guided

# Subsequent deploys
./scripts/deploy.sh

This deploys a Lambda function with:

• Lambda Web Adapter layer for SSE streaming
• Function URL (no API Gateway needed)
• RESPONSE_STREAM invoke mode for real-time streaming

Configuration

Variable	Default	Description
IMAGEKIT_API_BASE_URL	https://imagekit-public-mcp-tools.stlmcp.com	Upstream API base URL
IMAGEKIT_API_HOST	stage-ik-agent-service.imagekit.io	Host header for upstream
MCP_SERVER_NAME	imagekit-mcp-server	Server name in MCP protocol
MCP_SERVER_VERSION	1.0.0	Server version
LOG_LEVEL	INFO	Logging level

MCP Client Configuration

Claude Desktop

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "imagekit": {
      "url": "https://<your-function-url>/sse"
    }
  }
}

Cursor

Add to MCP settings:

{
  "mcpServers": {
    "imagekit": {
      "url": "https://<your-function-url>/sse"
    }
  }
}

Local (stdio via uv)

{
  "mcpServers": {
    "imagekit": {
      "command": "uv",
      "args": ["run", "python", "-m", "src.server"],
      "cwd": "/path/to/this/project"
    }
  }
}

Project Structure

├── src/
│   ├── __init__.py
│   ├── server.py           # MCP server + Starlette ASGI app
│   ├── client.py           # Async HTTP client for upstream API
│   ├── config.py           # Environment variable configuration
│   └── lambda_handler.py   # Lambda entry point (uvicorn)
├── tests/
├── scripts/
│   └── deploy.sh           # One-command SAM deployment
├── template.yaml           # AWS SAM template
├── samconfig.toml          # SAM deployment defaults
├── Dockerfile              # Lambda container image
├── pyproject.toml          # Project config + dependencies (uv)
├── .env.example            # Environment variable reference
└── openapi.json            # Upstream API spec (reference)