RILLA Shield exposes its Australian fraud-detection engine over the Model Context Protocol. Connect once from Claude, ChatGPT, Copilot or Cursor — your assistant can then check invoices, ABNs, BSBs, cousin domains and suspicious emails without you leaving the chat.
https://rillashield.app/mcpPublic endpoint · rate-limited per IP (5/hr anonymous, 100/hr with partner API key) · optionalX-MCP-API-Key header for white-label / referral partners. Streamable HTTP transport.
Five tools, one engine. Same detection logic that powers the RILLA Shield web app — every response is evidence and risk score, never a payment verdict.
check_invoiceFull 15-layer fraud engine on any invoice or payment email.
check_emailBusiness Email Compromise, phishing and bank-change screening.
check_abnValidates any ABN via the ATO mod-89 checksum.
check_bsbIdentifies the bank + flags fraud-prone BSB prefixes.
check_domainCousin/lookalike domain detector for BEC attacks.
After connecting, paste this into your assistant. It'll pick the right RILLA Shield tool automatically.
I just received this invoice from a supplier. Can you use RILLA Shield to check if it's genuine? From: accounts@bunnnings-trade.co Subject: Updated banking details — please pay latest invoice to new account Body: Hi mate, we've changed banks. Please pay invoice #4471 ($12,480 inc GST) into: BSB 802-985 Account 12345678 Account name: Bunnings Trade Supplies ABN 12 345 678 901 Urgent — needs to be paid today so we can release the timber order.
RILLA Shield returns risk scores and forensic evidence — it never says "safe to pay" or "approved". Final approval decisions rest with the human payer. Tools are rate-limited, stateless, and do not retain the content submitted through the public MCP endpoint.