Dr Urls speaks Model Context Protocol. An assistant can list your sites, read an audit, find issues, start a scan, and — when the answer needs a person rather than a crawler — hire one.
{
"mcpServers": {
"drurls": {
"url": "https://drurls.com/api/v1/mcp",
"headers": { "Authorization": "Bearer <your-api-key>" }
}
}
}Mint a key in the dashboard. Full tool list on the MCP page.
With an account — tools draw on your credit balance, exactly as the web app does. Without one — an unauthenticated call to a paid tool answers HTTP 402 describing what it accepts; pay and retry with an X-PAYMENT header. That second path exists because topping up credits needs a checkout, and a checkout needs a person — an agent that runs out at 3am would otherwise simply stop.
Reading is free: listing sites, reading a report, searching issues and checking scan status only surface data you already paid to produce. Work costs credits, at $0.10 per credit when paying by wallet.
| Action | Credits | By wallet |
|---|---|---|
| Scans | 1 | $0.10 |
| AI reviews | 2 | $0.20 |
| Tester orders | 1 | $0.10 |
Hiring a tester is priced by the offer, not by the action — see the tester catalogue. Machine-readable prices: /pricing.md.
An audit measures what is measurable. It cannot tell you whether a layout is broken on a real phone, whether a checkout confuses people, or what a screen reader actually announces. For those, hire_tester buys a real person on a real device, who returns a recorded walkthrough with a screenshot and the console output of every step.