Model Context Protocol (MCP) server that exposes The Metropolitan Museum of Art public collection via natural-language tools (list departments, search objects, retrieve object details with image).
https://github.com/mikechao/metmuseum-mcpTurn any AI assistant into an art research tool with direct access to The Met's 400,000+ object collection. This MCP server bridges The Metropolitan Museum of Art's public API with natural language queries, giving you structured art data and images without wrestling with REST endpoints.
You know the drill: find an interesting dataset, dig through API docs, write HTTP requests, parse responses, handle pagination. The Met's collection API is well-designed, but you still need to learn their object IDs, department structures, and search parameters.
This MCP server eliminates that friction. Ask your AI assistant "Show me Van Gogh paintings from the European department" and get back structured data with images automatically loaded as resources.
Department Discovery: Get all 21 Met departments without memorizing IDs
"Can you list all the departments at The Met?"
Smart Search: Query by title, artist, medium, or general terms with optional image filtering
"Find all objects with 'sunflower' in the title that have images"
"Show me Japanese ceramics from the Asian Art department"
Full Object Details: Complete metadata plus base64-encoded images added directly to your assistant's context
"Get details for Met object 436532"
// Returns: title, artist, bio, dimensions, medium, credit line, tags, and image
Art Research: Compare artistic styles across periods by pulling multiple works into a single conversation context. The assistant can analyze techniques, color palettes, and compositional elements across high-resolution images.
Educational Content: Generate lesson plans with actual artwork examples. Search for "portrait" + "Renaissance" and get authentic period pieces with provenance data.
Creative Inspiration: Pull random objects from specific departments to spark design ideas. The image resources let AI assistants provide visual analysis and suggestions.
Collection Analysis: Query trends across the collection - "How many works does The Met have from the 1920s?" or analyze representation across different cultures and time periods.
Add to your Claude Desktop config:
{
"mcp-servers": {
"met-museum": {
"command": "npx",
"args": ["-y", "metmuseum-mcp"]
}
}
}
No API keys, no rate limits, no authentication dance. The Met's public API is genuinely public, and this server respects that simplicity.
You could certainly hit the Met API directly, but this server handles the annoying parts:
The server's built with TypeScript, uses Zod for validation, and packages cleanly with the MCP SDK. It's the kind of tool that takes 2 minutes to add but saves hours of API wrangling when you need art data.
Perfect for developers building educational apps, content tools, or research platforms where art context matters. The Met's collection spans 5,000 years across every medium - having that accessible through natural language opens up workflows that weren't practical before.