Met Museum Mcp Server

---

Overview

The Metropolitan Museum of Art's public Collection API. Search the collection by keyword and filters, then fetch full object records — metadata, provenance, and CC0 open-access images — from any MCP client. Runs as a stdio process, a local Streamable HTTP server, or the public hosted endpoint above.

Tools

Tool	Description
met_list_departments	Return all 19 curatorial departments with their numeric IDs and display names
met_search_collections	Search the collection by keyword with filters for department, date range, medium, geography, on-view status, public-domain status, and highlight designation
met_get_object	Fetch full records for one or more object IDs — metadata, provenance, artist info, CC0 image URLs, tags, and Wikidata links

Capability reference

met_list_departments _tool

• No input required — returns all 19 curatorial departments in one call
• Each entry pairs departmentId (numeric) with displayName (e.g., "European Paintings", "Egyptian Art")
• departmentId values are the valid inputs for met_search_collections's departmentId filter

---

met_search_collections _tool

• Keyword q (required) matches title, artist name, culture, medium, tags, and other text fields; broad terms return large ID sets
• Filters: departmentId (validated against met_list_departments; an unrecognized ID is rejected), medium (maps to classification, not material — e.g. "Paintings", not "Oil on canvas"), dateBegin/dateEnd (integer years, negative = BCE, both required together), geoLocation (country/region/city, multiple values AND-combined), hasImages, isOnView
• isPublicDomain and isHighlight accept true only — the upstream index is unsound on the false arm, so omit the filter instead of passing false
• Every filter draws on a partial upstream index: a filtered search can omit objects whose own record satisfies the filter, so absence from the results is not evidence about an object — drop the filter to widen, and confirm per-object status with met_get_object
• A filtered search is cross-checked against the same query run unfiltered so results match the keyword; when that check can't complete in time (a keyword broad enough to time out on its own), the page returns unverified with a notice
• Paginate with limit (default 20, max 500) and offset (default 0); nextOffset continues where a page left off (null once exhausted), and offset >= total marks a page past the end of the result set rather than an exhausted query
• Returns total, returned, truncated, remaining, nextOffset, and the resolved offset; object IDs resolve to full records via met_get_object (up to 20 per call)
• Typed errors: no_results, invalid_date_range, invalid_filter (blank q, medium, or geoLocation), invalid_department, search_timeout — each carries a recovery hint

---

met_get_object _tool

• Accepts 1–20 IDs per call; a repeated ID is fetched and returned once, at its first position
• Partial-success batching — a 404 or fetch failure doesn't fail the whole call; failed[] reports per-ID error detail, and the call fails only when every ID fails (all_not_found / all_failed)
• Full record: metadata, provenance, artist/constituent data (Getty ULAN + Wikidata URLs), controlled-vocabulary tags (Getty AAT + Wikidata), a nine-field geography findspot block, and structured measurements — sparse fields are empty string or null, never fabricated
• Records are never truncated individually — a call whose combined records exceed a serialized-bytes budget returns fewer of them, listing the rest in deferred[] with sizes to re-request; content[] re-renders the admitted records, so the delivered response runs roughly twice the budget
• isPublicDomain/hasCC0Image gate image URLs — non-public-domain objects return empty primaryImage, primaryImageSmall, and additionalImages
• Canonical objectURL and per-object objectWikidata_URL for human follow-up and external enrichment

Features

Built on @cyanheads/mcp-ts-core: stdio and Streamable HTTP transports, pluggable auth (none / jwt / oauth), swappable storage (in-memory, filesystem, Supabase, Cloudflare KV/R2/D1), structured logging with optional OpenTelemetry tracing.

Met Museum-specific:

• 500K+ artworks spanning 5,000 years from the Met's public collection API
• CC0 open-access data from The Metropolitan Museum of Art — free to use without permission or attribution
• Parallel batch fetching with configurable concurrency for met_get_object
• Linked data on every object — Getty ULAN and AAT URLs, Wikidata entity URLs for artists, tags, and works

Agent-friendly output:

• Provenance on every record — isPublicDomain and hasCC0Image flags distinguish CC0 objects from works with inaccessible images, so agents can reason about what they can actually display
• Partial failure reporting — met_get_object returns objects and failed arrays so callers receive successful records alongside structured per-ID error context
• Truncation signaling — met_search_collections returns total, returned, truncated, remaining, nextOffset, and the resolved offset fields so agents know when to refine filters, increase limit, or page further with offset; offset >= total marks a page that is empty because the offset ran past the end rather than because the query is exhausted
• Byte-budget disclosure — met_get_object reports deferred[] records with their sizes when a batch exceeds its serialized-response budget, so callers can size a follow-up call precisely

Getting started

Public Hosted Instance

A public instance is available at https://met-museum.caseyjhand.com/mcp — no installation required. Point any MCP client at it via Streamable HTTP:

{
  "mcpServers": {
    "met-museum-mcp-server": {
      "type": "streamable-http",
      "url": "https://met-museum.caseyjhand.com/mcp"
    }
  }
}

Self-Hosted / Local

Add the following to your MCP client configuration file.

{
  "mcpServers": {
    "met-museum-mcp-server": {
      "type": "stdio",
      "command": "bunx",
      "args": ["@cyanheads/met-museum-mcp-server@latest"],
      "env": {
        "MCP_TRANSPORT_TYPE": "stdio",
        "MCP_LOG_LEVEL": "info"
      }
    }
  }
}

Or with npx (no Bun required):

{
  "mcpServers": {
    "met-museum-mcp-server": {
      "type": "stdio",
      "command": "npx",
      "args": ["-y", "@cyanheads/met-museum-mcp-server@latest"],
      "env": {
        "MCP_TRANSPORT_TYPE": "stdio",
        "MCP_LOG_LEVEL": "info"
      }
    }
  }
}

Or with Docker:

{
  "mcpServers": {
    "met-museum-mcp-server": {
      "type": "stdio",
      "command": "docker",
      "args": [
        "run", "-i", "--rm",
        "-e", "MCP_TRANSPORT_TYPE=stdio",
        "ghcr.io/cyanheads/met-museum-mcp-server:latest"
      ]
    }
  }
}

For Streamable HTTP, set the transport and start the server:

MCP_TRANSPORT_TYPE=http MCP_HTTP_PORT=3010 bun run start:http
# Server listens at http://localhost:3010/mcp

Prerequisites

• Bun v1.4.0 or higher (or Node.js v24+).
• No API key required — the Met Collection API is public and unauthenticated.

Installation

1. Clone the repository:

git clone https://github.com/cyanheads/met-museum-mcp-server.git

2. Navigate into the directory:

cd met-museum-mcp-server

3. Install dependencies:

bun install

4. Configure environment:

cp .env.example .env
# edit .env as needed (all vars are optional)

Configuration

All configuration is validated at startup via Zod schemas in src/config/server-config.ts.

Variable	Description	Default
MCP_TRANSPORT_TYPE	Transport: stdio or http	stdio
MCP_HTTP_PORT	HTTP server port	3010
MCP_SESSION_MODE	HTTP session mode: auto, stateful, or stateless. This server declares stateless in createApp(), so it applies whenever the variable is unset; an explicit value overrides it. (auto, the framework schema default, resolves to stateful.)	stateless
MCP_AUTH_MODE	Authentication: none, jwt, or oauth	none
MCP_LOG_LEVEL	Log level (debug, info, warning, error)	info
LOGS_DIR	Directory for log files (Node.js only)	<project-root>/logs
OTEL_ENABLED	Enable OpenTelemetry instrumentation	false
MET_BASE_URL	Met Collection API base URL (override for local stubs)	https://collectionapi.metmuseum.org/public/collection/v1
MET_REQUEST_TIMEOUT_MS	Per-request HTTP timeout in milliseconds	10000
MET_BATCH_CONCURRENCY	Max parallel fetches in met_get_object	5

See .env.example for the full list of optional overrides.

Running the server

Local development

• Build and run:

  # One-time build
  bun run rebuild
  
  # Run the built server
  bun run start:stdio
  # or
  bun run start:http
  

• Run checks and tests:

  bun run devcheck   # Lint, format, typecheck, security
  bun run test       # Vitest test suite
  bun run lint:mcp   # Validate MCP definitions against spec
  

Docker

docker build -t met-museum-mcp-server .
docker run --rm -p 3010:3010 met-museum-mcp-server

The Dockerfile defaults to HTTP transport, stateless session mode, and logs to /var/log/met-museum-mcp-server. OpenTelemetry peer dependencies are installed by default — build with --build-arg OTEL_ENABLED=false to omit them.

Project structure

Directory	Purpose
src/index.ts	createApp() entry point — registers tools and inits the Met service.
src/config	Server-specific environment variable parsing and validation with Zod.
src/mcp-server/tools	Tool definitions (*.tool.ts) — met_list_departments, met_search_collections, met_get_object.
src/services/met	Met Collection API client — HTTP, request timeout, response normalization.
tests/	Unit and integration tests mirroring src/.

Development guide

See CLAUDE.md for development guidelines and architectural rules. The short version:

• Handlers throw, framework catches — no try/catch in tool logic
• Use ctx.log for request-scoped logging, ctx.state for tenant-scoped storage
• Register new tools via the arrays in createApp() in src/index.ts
• Wrap external API calls: validate raw → normalize to domain type → return output schema; never fabricate missing fields

Contributing

Issues are welcome. Run checks and tests before submitting:

bun run devcheck
bun run test

Data attribution

Data from The Metropolitan Museum of Art Collection API (CC0).

License

Apache-2.0 — see LICENSE for details.