There is no aggregator re-shaping the data and no cache layer that can go stale, so what you read is what the government's system emits at the moment you ask. The trade is that you deal with each jurisdiction's own field names rather than a single normalized schema — which is exactly what makes the answer traceable back to the source record.
A hosted MCP endpoint that queries 30 official government company registries in real time — UK Companies House, France RNE, Spain BORME, Poland KRS, Korea OpenDART and more — and returns each registry's own response unmodified.
- Company search and profiles across 30 national registries — search_companies, list_jurisdictions
- Ownership: get_shareholders and get_persons_with_significant_control, chainable across jurisdictions
- Raw filings: fetch_document returns the government's own XHTML iXBRL, PDF or XBRL bytes
- live_company_accounts_xbrl for financial-statement data
- search_specialised_records for registries that do not expose the standard concepts
- Prebuilt workflows: ubo_cross_border_chain_walker, shell_company_detector, director_search_pep_screening, corporate_filing_monitor
- Identifiers preserved, so any response can be traced back to the government's own URL
Nothing to install and no API key to paste: the hosted endpoint at https://openregistry.sophymarine.com/mcp speaks Streamable HTTP and uses OAuth 2.1 with Dynamic Client Registration, so your client registers itself. The free tier allows 30 requests per minute across 6 core tools and every jurisdiction; paid keys raise quota and depth.
One command — claude mcp add --transport http openregistry https://openregistry.sophymarine.com/mcp