ShoppTrust for business and AI agents
The same evidence-based shop verdicts, for payment providers, banks, marketplaces, browsers and AI shopping agents. Every response carries the verdict, the reasons and the full evidence, so you can show your users why.
REST API
curl -H "Authorization: Bearer st_live_..." "https://shopptrust.com/api/v1/check?url=example.co.uk"
GET /api/v1/check?url=- verdict and evidence (cached up to 24 h, or a live check of 20-60 s;max_age_hours,refresh)GET /api/v1/report/{domain}- the latest stored report, never a live checkPOST /api/v1/batch- up to 50 shops; pollGET /api/v1/batch/{job}lang=heon any call returns the headline, reasons and evidence in Hebrew (the response fields stay the same).
Interactive reference: /api/docs. Each check costs one call of your monthly quota; the remaining
quota is in the X-Quota-Remaining header.
AI agents (MCP)
Agents connect to /mcp (streamable HTTP, same API key) and get two tools: check_shop(url)
and get_shop_report(domain), both with an optional lang (en or he). Local assistants can run python -m shoptrust mcp over stdio.
Verdicts
TRUSTED · KNOWN_WITH_ISSUES (a known business with some issues) · NOT_ENOUGH_EVIDENCE (not a safety rating) · SUSPICIOUS · SCAM · NOT_A_SHOP (the site does not sell online) -
see how verdicts are decided. Only TRUSTED is a recommendation to buy: the recommended field is true for it alone.