companies-house-mcp
Enables querying UK company data including search, profiles, officers, filings, and persons with significant control via Companies House API.
README
companies-house-mcp
Production-grade Model Context Protocol server for the UK Companies House API.
Point any MCP-aware assistant (Claude Desktop, Claude Code, Cursor, custom agents) at UK company data. Auth, rate limiting, in-memory caching, structured errors, and a real test suite included — not a tutorial demo.
Status: alpha (v0.1). Five core tools shipped. Free filing-history documents, resumable-filings, and disqualified-officers lookups on the roadmap.
Why this exists
The official MCP reference servers are explicitly educational. They ship without auth, rate limiting, caching, or tests — deliberately, so learners can read them in an afternoon. Every serious team building on MCP ends up re-implementing the same production concerns from scratch.
This is one of them, done properly, against a UK data source that accountants, KYC/AML teams, sales-ops, and journalists genuinely use every day.
Tools exposed
| Tool | Description |
|---|---|
search_companies |
Free-text search across all registered UK companies. |
get_company_profile |
Full profile for a company by number (name, status, address, SIC codes, dates, accounts). |
list_officers |
Directors, secretaries, and LLP members for a company. |
list_filings |
Filing history (annual accounts, confirmation statements, appointments, resolutions). |
get_psc |
Persons with significant control (25%+ ownership, voting rights, etc.). |
Every tool returns structured JSON — the client model doesn't have to parse HTML or free-form text.
Install & configure
Requires Node.js 20+ and a free Companies House API key from developer.company-information.service.gov.uk.
Claude Desktop
Edit your claude_desktop_config.json:
{
"mcpServers": {
"companies-house": {
"command": "npx",
"args": ["-y", "companies-house-mcp"],
"env": {
"CH_API_KEY": "your-key-here"
}
}
}
}
Restart Claude Desktop. Ask it: "Look up MORRIS AI LTD on Companies House and tell me who the directors are."
Claude Code
claude mcp add companies-house npx -y companies-house-mcp -- --env CH_API_KEY=your-key-here
Local dev
git clone https://github.com/shadowborn-dev/companies-house-mcp
cd companies-house-mcp
npm install
cp .env.example .env # add your API key
npm run build
npm start
Docker
docker build -t companies-house-mcp .
docker run --rm -i -e CH_API_KEY=your-key-here companies-house-mcp
Try it in the browser (MCP Inspector)
The fastest way to poke at the tools without wiring the server into Claude Desktop is the official MCP Inspector — a web UI that connects to any MCP server and lets you call its tools with a form.
# put your key in .env first (or export CH_API_KEY=... in the shell)
npm run inspect
That builds the server, launches the Inspector in a local browser tab, and connects to the built server over stdio. You'll see the five tools listed on the left; click one, fill in the form on the right (company_number: "12345678" for get_company_profile, or query: "morris" for search_companies), hit Run tool, and the raw JSON response comes back below. Handy for verifying a tool works before you plug the server into Claude.
Production concerns handled
Auth. Companies House uses HTTP Basic with the API key as the username. This server wraps that transparently — you only ever set CH_API_KEY.
Rate limiting. Companies House enforces 600 requests per 5-minute rolling window per API key. This server ships a token-bucket limiter tuned to match — requests over the budget queue and wait rather than being sent through and 429'd back. Configurable via CH_RATE_LIMIT_MAX and CH_RATE_LIMIT_WINDOW_SECONDS.
Caching. In-memory LRU with per-entry TTL (default 5 minutes). Company profiles rarely change intraday, so a cache dramatically cuts request count for repeated lookups. Configurable via CH_CACHE_TTL_SECONDS and CH_CACHE_MAX_ENTRIES.
Structured errors. Every failure surfaces as a typed error class the model can reason about — NotFoundError, UnauthorizedError, RateLimitedError, UpstreamError — instead of a generic 500.
Tests. Vitest covers the rate limiter, cache eviction, and the HTTP client with mocked fetch. npm test runs the whole suite.
Configuration reference
| Env var | Default | Purpose |
|---|---|---|
CH_API_KEY |
(required) | Companies House API key. |
CH_RATE_LIMIT_MAX |
600 |
Max requests per window. |
CH_RATE_LIMIT_WINDOW_SECONDS |
300 |
Rolling window in seconds. |
CH_CACHE_TTL_SECONDS |
300 |
Cache entry TTL. |
CH_CACHE_MAX_ENTRIES |
1000 |
Max cache size before LRU eviction. |
Roadmap
- v0.2 —
get_filing_document(fetches the PDF or iXBRL underlying a filing). - v0.3 —
disqualified_officerssearch +officer_appointments(all companies a person is an officer of). - v0.4 — Streamable HTTP transport for remote/hosted deployments.
- v0.5 — Optional Redis-backed cache + rate limiter for multi-instance deployments.
Honest caveats
- The Companies House API returns data as filed — occasionally with typos, inconsistent capitalisation, or delays of days between filing and appearing. This server does not clean or reconcile any of that.
- Rate-limit budgeting is per-API-key. If you share a key across multiple instances of this server, they will not coordinate — colocate to one instance or supply a distributed limiter in v0.5.
- Not affiliated with or endorsed by Companies House.
License
MIT. See LICENSE.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.