What I learned shipping a pay‑per‑call Irish rent API for AI agents

Description and use

A pay‑per‑call API for Irish rent and property data was built using x402 and MPP payment challenges, made discoverable via MCP, and logged privacy‑safe analytics.

Problem

Provide a low‑cost, account‑free API for Irish rent and property data that can be paid for per call by AI agents.

Outcome

The API was successfully deployed with payment challenges and discoverability, but had not yet attracted paid usage at the time of reporting.

Workflow

• 1. Design API endpoints for rent and price data.
• 2. Implement payment challenges: x402 (USDC via CDP facilitator) and MPP (Tempo).
• 3. Serve HTTP 402 before input validation to block crawlers.
• 4. Provide /openapi.json with x-payment-info for x402scan and a real MPP challenge for mppscan.
• 5. Test end‑to‑end settlement on testnet before switching to mainnet.
• 6. Publish API as a stateless MCP server and register it in the official MCP registry using HTTP domain auth.
• 7. Add llms.txt, server card, and /.well-known/x402 for agent discovery.
• 8. Log per‑route analytics with HMAC‑based caller IDs to preserve privacy.
• 9. Monitor usage and revenue (currently 0 paid calls).

Evidence

An HTTP API over Irish public data: average asking rent by county, town or Dublin postal district (CSO/RTB RIQ02, CC BY 4.0), a summary of the current rent‑increase rules, and county house‑price aggregates from the PSRA Property Price Register (median/p25/p75, groups under 5 sales suppressed, no addresses stored).

Source