
Met Museum MCP Server
Search and explore The Metropolitan Museum of Art collection, including artwork details and images, via natural language.
Add to your client
Copy the config for your MCP client and paste it into its config file.
npx -y metmuseum-mcpPaste into ~/Library/Application Support/Claude/claude_desktop_config.json
{
"mcpServers": {
"met-museum-mcp-server": {
"command": "npx",
"args": [
"-y",
"metmuseum-mcp"
]
}
}
}Step-by-step guides: Add to Claude Desktop · Add to Cursor · Add to Windsurf
Before you start
- Node.js v18 or later
About Met Museum MCP Server
Met Museum MCP Server exposes The Metropolitan Museum of Art's open-access collection to AI assistants. Through four tools, models can list The Met's departments, search for objects (filtering by images, title, department, and paginated results), and retrieve complete open-access data for a specific object — including its image returned as a base64-encoded JPEG content block. An Open Met Explorer MCP App provides an interactive UI for browsing live search results inside clients that support MCP Apps. It runs via npx with no install or credentials, over stdio by default or Streamable HTTP with the --http flag.
Tools & capabilities (4)
list-departmentsLists all the valid departments at The Met, each with a Department ID and display name. Takes no inputs.
search-museum-objectsSearches for objects in The Met by a query term. Optional filters: hasImages (only objects with images), title (match against the title field only), departmentId (restrict to a department), and paginated results via page and pageSize (max 100). Returns the total count and a list of matching object IDs.
get-museum-objectRetrieves a specific object by objectId with all open-access data (title, artist, bio, department, credit line, medium, dimensions, primary image URL, tags). If returnImage is true (default) and an image is available under Open Access, the image is included as a base64-encoded JPEG content block.
open-met-explorerLaunches the interactive Met Explorer MCP App (ui://met/explorer.html) to search, filter, and inspect objects from within MCP clients that support Apps. All inputs optional: q (initial search term), hasImages (default true), title (search titles only), departmentId (pre-select a department filter).
What this server can do
Met Museum MCP Server provides tools for these capabilities — tap one to see every MCP server that does the same:
When to use it
- Explore the works of a specific artist, e.g. Vincent van Gogh, across The Met's collection
- Find a particular artwork by title, such as 'Corridor in the Asylum'
- Browse paintings from a specific department like Asian Art
- Search for art by theme or subject, e.g. anything with 'cat' in the title
- Retrieve detailed metadata and open-access images for a known object ID
Security notes
No API key or authentication is required; the server calls the public Met Collection API anonymously. When run with the Streamable HTTP transport (--http), the server binds to 127.0.0.1 by default and validates the Host header against an allowlist (ALLOWED_HOSTS, default localhost,127.0.0.1); review HOST/PORT/ALLOWED_HOSTS before exposing it beyond localhost.
Met Museum MCP Server FAQ
Do I need an API key or account?
No. The server uses the public Met Collection API anonymously, so no API key or authentication is required.
Is this an official Metropolitan Museum of Art project?
No. It is a third-party implementation of the public Met Collection API and is not officially associated with The Metropolitan Museum of Art.
Can I use it with ChatGPT or other HTTP clients?
Yes. Run the server with the --http flag (npx -y metmuseum-mcp --http) to expose a Streamable HTTP MCP endpoint at http://localhost:3001/mcp, then connect it as a connector (for ChatGPT, typically via a tunnel such as ngrok).
Are artwork images returned?
Yes. get-museum-object returns the object's image as a base64-encoded JPEG content block when returnImage is true (the default) and the image is available under Open Access.
Alternatives to Met Museum MCP Server
Compare all alternatives →Official MCP reference server that fetches a URL and returns its content as clean Markdown, with chunking.
Official Firecrawl MCP server — scrape, crawl, map, search, and structured extraction for any LLM client.
Exa's neural web search and crawling MCP server — runs locally via npx or as a hosted remote endpoint.
Compare Met Museum MCP Server with: