WIP: Local MCP Server Config

[Gabe-Levin]

Add Space2Stats MCP Server

What I changed

• Added a new space2stats_mcp/ package — an MCP (Model Context Protocol) server that exposes the Space2Stats API as tools for AI assistants (Claude Desktop, Claude Code, Cursor, VS Code Copilot, etc.)
• 11 tools covering data discovery (list_fields, list_topics), admin boundary fetching (fetch_admin_boundaries), and spatial queries (get_summary, get_aggregate, get_timeseries) plus hex-level variants
• STAC catalog integration to enrich results with field descriptions, topics, and source metadata
• .mcp.json at the repo root for automatic server configuration in supported IDEs
• README with installation, configuration for 5+ MCP clients, and testing instructions

How to test it

1. Install the package: pip install -e space2stats_mcp/
2. Start the server: space2stats-mcp (should start without errors, waiting for JSON-RPC input — Ctrl+C to stop)
3. Verify tools work:
   • list_fields → returns 106 fields
   • fetch_admin_boundaries with {"iso3": "AND", "adm": "ADM0"} → returns GeoJSON
   • get_aggregate with a small AOI → returns aggregated statistics
4. Or test in Claude Code: open the repo, confirm the MCP connects, and ask "What population fields are available in Space2Stats?"

[github-actions]

PR Deployment Details:
🚀 PR deployed to https://r5kh8bh6m4.execute-api.us-east-1.amazonaws.com/