How it works
URL in. Score out.
An autonomous analyst behind every link — with controls you tune to your own risk appetite.
01
URL in
Send a single URL to /scan or up to 100 at once to /batch.
02
Proprietary algorithm
Our agentic model inspects the domain, content, redirects and behavior of each link.
03
Score out
Get a risk score from 0 to 1, a verdict and the evidence behind it.
Whitelists & blacklists
Your rules win
Add trusted domains to your whitelist so they always pass, and known-bad domains to your blacklist so they're always blocked — no model call needed.
Setting the temperature
Tune how strict it is
Lower temperature means the model only flags what it's very sure about. Raise it to catch more borderline links at the cost of more warnings.
POST /scan
{
"url": "https://example.com/login",
"temperature": 0.3,
"whitelist": ["mycompany.com"],
"blacklist": ["known-bad.net"]
}Level 2 API — /sandbox
Open it in a sandbox. Verify it.
When a link needs more than a score, send it to the Level 2 API. Each URL is opened in an isolated sandbox and its live behavior is verified before the verdict comes back. It's slower than /scan, as expected — so route only the URLs you choose, such as borderline results, to the sandbox.
$0.05 per sandbox scan, on every plan except Docker.
POST /sandbox
{
"url": "http://evil.co/1"
}Trust & security
Built to be trusted
TLS/HTTPS Encryption in Transit
All network traffic and API requests are strictly encrypted in transit using industry-standard TLS 1.3 to guarantee data privacy and integrity end-to-end.
Minimal retention
Scanned URLs are not retained on paid accounts with only usage statistics measured.
Compliance
SOC 2 Type II and GDPR programs, with DPAs available on request.
Responsible disclosure
We welcome reports from researchers via security@smarty-pants.ai.
Start scanning URLs in minutes
1,000 free checks every month. No credit card required.
Get your API key