starwell-mcp

Starwell is the verified data layer for AI: official statistics from Statistics Canada, FRED, the U.S. Bureau of Labor Statistics, the World Bank, SEC EDGAR, the U.S. Treasury, the European Central Bank, the UK Office for National Statistics, INSEE (France), the Australian Bureau of Statistics, the OECD, Eurostat, INE (Spain), the Czech National Bank, ILOSTAT, Statistics Netherlands, the CDC, Statistics Sweden, Statistics Norway, Statistics Finland, Statistics Austria, the Department of Statistics Malaysia, the U.S. Bureau of Economic Analysis, the Central Statistics Office of Ireland, Statistics Denmark, Statistics Poland, Statistics Estonia, Statistics Iceland, and the U.S. Bureau of Transportation Statistics, served as MCP tools. Every observation carries provenance, every series carries a verification status backed by golden-value checks against the live source, and the answer tool returns computed results cited to the official table.

This package is a thin stdio bridge to the Starwell remote MCP server. The tool surface always matches the server, so new sources and tools appear without a package update.

Tools (12): search_catalog, list_sources, list_datasets, get_series, get_series_stats, get_observations, factory_status, answer, deep_analysis, create_monitor, list_monitors, delete_monitor. Reads are open; answer, deep_analysis, and the monitor tools need a free API key (STARWELL_API_KEY).

Install per client

Remote-capable clients can skip this package and add the URL directly: https://starwell.dev/api/starwell/mcp (Streamable HTTP).

Claude Code

claude mcp add --transport http starwell https://starwell.dev/api/starwell/mcp
# or via stdio:
claude mcp add starwell -- npx -y starwell-mcp

Claude Desktop / claude.ai

Settings, Connectors, Add custom connector, URL https://starwell.dev/api/starwell/mcp. On plans without remote connectors, use the desktop config:

{
  "mcpServers": {
    "starwell": { "command": "npx", "args": ["-y", "starwell-mcp"] }
  }
}

Cursor

.cursor/mcp.json:

{
  "mcpServers": {
    "starwell": { "command": "npx", "args": ["-y", "starwell-mcp"] }
  }
}

OpenAI Codex CLI

~/.codex/config.toml:

[mcp_servers.starwell]
command = "npx"
args = ["-y", "starwell-mcp"]

Gemini CLI

~/.gemini/settings.json:

{
  "mcpServers": {
    "starwell": { "httpUrl": "https://starwell.dev/api/starwell/mcp" }
  }
}

ChatGPT (developer mode)

Settings, Apps, enable developer mode, add connector with the remote URL above.

Configuration

• STARWELL_MCP_URL: override the endpoint (staging, self-hosted).
• STARWELL_API_KEY: Bearer key from https://starwell.dev/account (free). Data reads work without one; answer, deep_analysis, and monitors require it.

Data licensing

Only openly licensed official sources are served, each with its license and attribution in every response. The set covers U.S. federal statistical works (public domain), the Statistics Canada Open Licence, World Bank CC BY-4.0, and the respective open licences of the ECB, OECD, ONS, INSEE, and ABS. This product uses the FRED API but is not endorsed or certified by the Federal Reserve Bank of St. Louis.

Docs: https://starwell.dev/docs

---

<!-- generated: scripts/sync-mcp-mirror.ts in the Starwell monorepo. Edit there, not here. --> This repository is the source of the starwell-mcp npm package. It is generated from Starwell's main repository, so pull requests here may be applied upstream by hand — but issues are read and very welcome.