Black Flag Alert MCP Server

UK company credit-risk intelligence as native tools for your AI agent — Claude, Cursor, Windsurf, Cline, ChatGPT connectors, Gemini, or any Model Context Protocol client.

Black Flag Alert scores every company registered in England & Wales with the R-Score (0–100) — filed accounts, filing behaviour, charges, CCJs and winding-up activity combined into one number — and serves it through one MCP endpoint.

• Endpoint: https://blackflagalert.com/api/mcp (streamable HTTP, MCP 2025-06-18)
• API keys: free trial, no card — https://blackflagalert.com/developers
• Coverage: England & Wales (5.1M companies). SC/NI/R/FC/OE prefixes are out of coverage.
• Methodology: https://blackflagalert.com/methodology

Tools

Connect (remote, no install)

Works with any client that supports remote MCP servers:

{
  "mcpServers": {
    "black-flag-alert": {
      "type": "http",
      "url": "https://blackflagalert.com/api/mcp",
      "headers": {
        "Authorization": "Bearer bfa_live_xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx"
      }
    }
  }
}

ChatGPT: Settings → Connectors → Create (paste the URL). Claude web: Settings → Connectors → Add custom. Claude Code: claude mcp add --transport http black-flag-alert https://blackflagalert.com/api/mcp --header "Authorization: Bearer <key>". Gemini: Settings → Apps → Connections.

Install locally (stdio)

uv tool install bfa-mcp        # once published to PyPI
# or from source:
uv sync                        # in this repo

{
  "mcpServers": {
    "black-flag-alert": {
      "command": "bfa-mcp",
      "env": { "BFA_API_KEY": "bfa_live_..." }
    }
  }
}

Environment variables

Example

You: "Should we extend £20k credit to Kimmeridge Projects?"

Agent: search_companies("Kimmeridge Projects") → 04466987 → get_company("04466987") → "R-Score 93.87 (Very Low risk), no CCJs, no outstanding charges, net assets £254K. Suggested exposure: £24,000."

License

MIT