Smithsonian Open Access MCP Server

A Model Context Protocol (MCP) server that provides AI assistants with access to the Smithsonian Institution's Open Access collections. This server allows AI tools like Claude Desktop to search, explore, and analyze over 3 million collection objects from America's national museums.

Quick Start

Option 1: npm/npx Installation (Easiest)

The npm package includes automatic Python dependency management and works across platforms:

# Install globally
npm install -g @molanojustin/smithsonian-mcp

# Or run directly with npx (no installation needed)
npx -y @molanojustin/smithsonian-mcp

# Set your API key
export SMITHSONIAN_API_KEY=your_key_here

# Start the server
smithsonian-mcp

Option 2: Automated Setup (Recommended for Python users)

The enhanced setup script now includes:

• ✅ API key validation - Tests your key before saving
• ✅ Service installation - Auto-install as system service
• ✅ Claude Desktop config - Automatic configuration
• ✅ Health checks - Verify everything works macOS/Linux:

chmod +x config/setup.sh
config/setup.sh

Windows:

config\setup.ps1

Option 3: Manual Setup

1. Get API Key: api.data.gov/signup (free)
2. Install: uv pip install -r config/requirements.txt
3. Configure: Copy .env.example to .env and set your API key
4. Test: python examples/test-api-connection.py

Verify Setup

Run the verification script to check your installation:

python scripts/verify-setup.py

Features

Core Functionality

• Search Collections: 3+ million objects across 24 Smithsonian museums
• Object Details: Complete metadata, descriptions, and provenance
• On-View Status - Find objects currently on physical exhibit
• Image Access: High-resolution images (CC0 licensed when available)
• Museum Information: Browse all Smithsonian institutions
• Collection Statistics: Comprehensive metrics with per-museum breakdowns (sampling-based estimates)

AI Integration

• 16 MCP Tools: Smart discovery, comprehensive search, museum-specific queries, exhibition status, contextual data access, and proactive collection type discovery
• Proactive Discovery: New tools help AI assistants understand API scope and available object types before searching, preventing confusion about archival vs. museum materials
• Smart Context: Contextual data sources for AI assistants including enhanced statistics
• Rich Metadata: Complete object information and exhibition details
• Exhibition Planning - Tools to find and explore currently exhibited objects
• Collection Analytics: Per-museum statistics with sampling-based accuracy
• Multi-Model Compatible: Works well with both advanced and simpler AI models through simplified tool interfaces

URL Validation & Anti-Guessing

• Easiest Solution: Use search_and_get_first_url() for one-step search + validated URL retrieval
• Mandatory Tool Usage: LLM must use get_object_url() tool for any URL retrieval - manual construction fails due to case sensitivity
• Flexible Identifiers: Supports Accession Numbers (F1900.47), Record IDs (fsg_F1900.47), and Internal IDs (ld1-...)
• URL Validation: Automatically selects authoritative record_link over API identifiers, handles case sensitivity

Integration

Claude Desktop

Option 1: Using npm/npx (Recommended)

1. Configure (claude_desktop_config.json):

{
  "mcpServers": {
    "smithsonian_open_access": {
      "command": "npx",
      "args": ["-y", "@molanojustin/smithsonian-mcp"],
      "env": {
        "SMITHSONIAN_API_KEY": "your_key_here"
      }
    }
  }
}

Option 2: Using Python installation

1. Configure (claude_desktop_config.json):

{
  "mcpServers": {
    "smithsonian_open_access": {
      "command": "python",
      "args": ["-m", "smithsonian_mcp.server"],
      "env": {
        "SMITHSONIAN_API_KEY": "your_key_here"
      }
    }
  }
}

2. Test: Ask Claude "What Smithsonian museums are available?"

mcpo Integration (MCP Orchestrator)

mcpo is an MCP orchestrator that converts multiple MCP servers into OpenAPI/HTTP endpoints, ideal for combining multiple services into a single systemd service.

Installation

# Install mcpo
uvx mcpo

# Or using uvx
uvx mcpo --help

Configuration

Create a examples/mcpo-config.json file:

{
  "mcpServers": {
    "smithsonian_open_access": {
      "command": "python",
      "args": ["-m", "smithsonian_mcp.main"],
      "env": {
        "SMITHSONIAN_API_KEY": "your_api_key_here"
      }
    },
    "memory": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-memory"]
    },
    "time": {
      "command": "uvx",
      "args": ["mcp-server-time", "--local-timezone=America/New_York"]
    }
  }
}

Running with mcpo

# Start mcpo with hot-reload
mcpo --config examples/mcpo-config.json --port 8000 --hot-reload

# With API key authentication
mcpo --config examples/mcpo-config.json --port 8000 --api-key "your_secret_key"

# Access endpoints:
# - Smithsonian: http://localhost:8000/smithsonian_open_access
# - Memory: http://localhost:8000/memory
# - Time: http://localhost:8000/time
# - API docs: http://localhost:8000/docs

Systemd Service

Create /etc/systemd/system/mcpo.service:

[Unit]
Description=MCP Orchestrator Service
After=network.target

[Service]
Type=simple
User=your-user
WorkingDirectory=/path/to/your/config
Environment=PATH=/path/to/venv/bin
ExecStart=/path/to/venv/bin/mcpo --config examples/mcpo-config.json --port 8000
Restart=always
RestartSec=10

[Install]
WantedBy=multi-user.target

# Enable and start service
sudo systemctl enable mcpo
sudo systemctl start mcpo
sudo systemctl status mcpo

Troubleshooting mcpo

See TROUBLESHOOTING.md for detailed mcpo troubleshooting, including:

• ModuleNotFoundError solutions
• Connection closed errors
• Port conflicts
• Path configuration issues

VS Code

1. Open Workspace: code .vscode/smithsonian-mcp-workspace.code-workspace
2. Run Tasks: Debug, test, and develop the MCP server
3. Claude Code: AI-assisted development with Smithsonian data

Available Data

• 19 Museums: NMNH, NPG, SAAM, NASM, NMAH, and more
• 3+ Million Objects: Digitized collection items
• CC0 Content: Public domain materials for commercial use
• Rich Metadata: Creators, dates, materials, dimensions
• High-Resolution Images: Professional photography

Data Accuracy & Sampling

Collection statistics for objects with images use sampling methodology to provide accurate estimates:

• Sample Size: Up to 1000 objects per query for statistical significance
• Methodology: Counts actual returned objects instead of relying on potentially buggy API totals
• Coverage: Includes per-museum breakdowns with individual sampling for each institution
• Transparency: All sampled counts are clearly marked as "(est.)" in outputs

This approach ensures reliable metrics while respecting API rate limits and avoiding the Smithsonian API's rowCount filtering bug.

Current API Limitations

Image URLs Not Available: The Smithsonian Open Access API currently does not provide image URLs or media data in detailed content responses. While the search API can filter objects by media type (e.g., online_media_type:Images), the actual image URLs are not included in the detailed object data returned by the content API. This appears to be a change in the API since the available documentation was published.

• Objects will show as having 0 images even when filtered for image content
• Image statistics are estimates based on search filtering, not actual media availability
• The system gracefully handles this limitation and continues to provide all other metadata

API Scope: Diverse Museum Collections: The Smithsonian Open Access API provides access to diverse collections across 24 Smithsonian museums, with each museum having distinct object types reflecting their unique focus areas. The discovery tools now correctly identify museum-specific collections with comprehensive object type intelligence gathered through systematic sampling.

• SAAM (American Art): Paintings, decorative arts, sculptures, drawings
• NASM (Air & Space): Aircraft, avionics, spacecraft, aviation equipment
• NMAH (American History): Historical artifacts, inventions, cultural objects
• CHNDM (Design Museum): Design objects, textiles, furniture, graphics
• Use discovery tools (get_museum_collection_types, check_museum_has_object_type) to explore available collections
• Each museum's collection reflects its institutional mission and expertise

MCP Tools

Search & Discovery