Give AI agents a trusted map of the web
Before an agent logs in, pays, submits a form, or uploads a document, Trustiry tells it whether the destination is real, official, and safe to act on. Available as a REST API and a native MCP server.
Agent-Safety API
One call, before any sensitive action.
GET https://trustiry.com/api/agent/check
?url=amaz0n-secure-login.com
&action=payment
Response
{
"verdict": "likely_impersonation",
"trust_score": 0,
"risk_score": 82,
"safe_to_proceed": false,
"blocked_actions": ["login", "payment", "document_upload"],
"official_alternative": "https://www.amazon.com",
"reason": "This website appears to imitate Amazon. Use the official website instead."
}
action can be login, payment, document_upload, submit_form, or create_account. For metered/production use, issue an API key at the developer dashboard and call /api/v1/agent-safety.
MCP server
Native Model Context Protocol tools for any MCP client (Claude Desktop, agents, IDEs).
{
"mcpServers": {
"trustiry": {
"command": "node",
"args": ["/opt/trustiry/apps/trustiry-mcp/dist/index.js"],
"env": { "TRUSTIRY_API_BASE": "https://trustiry.com/api" }
}
}
}
Tools exposed
check_urlIs a link real, fake, risky, or impersonating a brand?
agent_safetyIs it safe to login / pay / upload on this URL? Returns safe_to_proceed + blocked_actions.
find_officialThe verified official website for a brand/org (+ known fake clones).
website_trustFull trust profile + Schema.org/Trustiry data for a site.
Why agents need this
๐
Before login
Confirm it's the real portal, not a clone harvesting credentials.
๐ณ
Before payment
Block payment on impersonation / new / high-risk domains.
๐
Before upload
Don't send documents to unverified destinations.