Skip to main content
Glama

eu-us-opendata-mcp

MCP Stdio Tests License: MIT

Zero-dependency, zero-API-key Model Context Protocol (MCP) stdio server in pure Python 3.10+ for querying 4 official European & U.S. government open-data registries directly from Claude Desktop, Claude Code, Cursor, Windsurf, Google Antigravity, or Codex.


Included MCP Tools (100% Free Official APIs, Zero Keys)

Tool

Official Upstream Registry

Key Fields Returned

search_french_companies

DINUM / INSEE / INPI (recherche-entreprises.api.gouv.fr)

SIREN, computed EU VAT (FRxx...), NAF/APE, city/ZIP, Qualiopi/RGE filters, Annuaire & Pappers URLs

search_eu_public_tenders

EU TED Search API v3 (api.ted.europa.eu/v3 — 27 EU countries)

Tender ID, title, buyer name, buyer email, phone & website, CPV codes, deadline & official notice URL

search_clinical_trials

U.S. NIH ClinicalTrials.gov REST API v2 (clinicaltrials.gov/api/v2)

NCT ID, title, status, phases, lead sponsor, enrollment, central contact email/phone & study URL

search_us_npi_providers

U.S. CMS NPPES NPI Registry API v2.1 (npiregistry.cms.hhs.gov)

10-digit NPI, doctor/clinic name, primary specialty, city/state, direct phone & fax, NPPES profile URL


Related MCP server: mcp-datagouv

Quick Install & Run (No API Keys Required)

Option 1 — Run directly with uvx (recommended)

{
  "mcpServers": {
    "eu-us-opendata": {
      "command": "uvx",
      "args": [
        "--from",
        "git+https://github.com/yannabadie/eu-us-opendata-mcp.git",
        "eu-us-opendata-mcp"
      ]
    }
  }
}

Option 2 — Install with pip

pip install git+https://github.com/yannabadie/eu-us-opendata-mcp.git
eu-us-opendata-mcp

Option 3 — Run from source (zero dependencies)

git clone https://github.com/yannabadie/eu-us-opendata-mcp.git
cd eu-us-opendata-mcp
PYTHONPATH=src python3 -m eu_us_opendata_mcp.server

Run Tests

PYTHONDONTWRITEBYTECODE=1 python3 -m unittest discover -s tests -v

Need Bulk Cloud Exports (1,000–10,000+ Records), Daily n8n Alerts, or Hosted Remote MCP?

This local MCP server is built for fast interactive agent lookups (up to 25 results per tool call). For bulk CSV/Excel/JSON extraction, scheduled daily monitoring, BODACC insolvency KYC screening, BOAMP French tenders, award winner intelligence, and ready-to-import n8n workflows, use our hosted Pay-Per-Event Apify Actors (free $5/mo Apify credits included):


License

MIT

Available Tools

4 tools
search_clinical_trialsA

Search 500,000+ clinical studies via official U.S. NIH ClinicalTrials.gov REST API v2 (returns NCT ID, title, status, phases, lead sponsor, enrollment, central contact email/phone & study URL).

Inputs: limit, status, sponsor, location, +2 more.

search_eu_public_tendersA

Search active European Union public procurement tenders across 27 EU countries via official EU TED Search API v3 (returns title, buyer name, buyer email/phone/website, CPV codes, deadline & official URL).

Inputs: limit, country, keyword, days_back, +1 more.

search_french_companiesC

Search official French companies via DINUM/INSEE/INPI SIRENE/RNE Open Data API (15M+ entities: SIREN, SIRET, computed EU VAT, NAF code, revenue, labels, address & Pappers link).

Inputs: limit, query, est_rge, naf_code, +3 more.

search_us_npi_providersA

Search 8,000,000+ U.S. physicians, dentists, clinics, hospitals & pharmacies via official U.S. CMS NPPES NPI Registry API v2.1 (returns NPI, specialty, practice phone/fax, address & authorized official).

Inputs: city, limit, state, specialty, +2 more.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    A3.6/5.0

    Scored across 4 tools

    Disambiguation5/5

    Each tool targets a completely separate data domain (French company registry, EU TED tenders, NIH clinical trials, CMS NPI registry), so an agent can pick the right one from the entity type alone. There is no overlap in purpose or return shape between any two tools.

    Naming Consistency5/5

    All four tools follow an identical search_<domain_entity> snake_case pattern, with the geographic/domain qualifier consistently placed first. The pattern is fully predictable and readable.

    Tool Count4/5

    Four tools is lean but each maps to a distinct authoritative dataset, so none feels redundant. It is on the thin side for a multi-source open-data server, since there are no companion detail/lookup tools.

    Completeness3/5

    The surface is search-only: there is no get-by-identifier tool (e.g. fetch company by SIREN, tender by notice ID, provider by NPI) and no way to page through or expand results. Search covers each domain broadly, but agents hit a dead end when they need full record detail or follow-up retrieval.

    Maintenance

    ActivityMaintained
    ResponsivenessNo recent community issues

    Related MCP Connectors

    Related MCP Servers