How Scantinel Works
How Scantinel works — scanners, then AI analysis, then AI pen testing
Proven scanners map your site and run the checks. Frontier AI models then triage every result, and on Premium an AI agent pen tests the endpoints the scan discovered. Every step is non-destructive and ownership-gated.
Ownership verification
Before any active probe runs, Scantinel verifies you own the target. This protects third parties from unauthorized scanning and gives you a documented record that the assessment was authorized.
DNS TXT record
Add a scantinel-verify= TXT record to your domain. Confirmed within seconds via DNS resolution.
HTTP file upload
Place a small verification file at /.well-known/scantinel-verify.txt. Confirmed via HTTP fetch.
HTML meta tag
Add a <meta name="scantinel-verify"> tag to your homepage. Confirmed when Scantinel fetches your page.
Email challenge
Scantinel sends a verification email to admin@, security@, or webmaster@ at your domain. Click to confirm.
What gets checked, by depth
| Check category | Essential | Full | Premium |
|---|---|---|---|
| DNS records audit | ✓ | ✓ | ✓ |
| TLS/SSL configuration | ✓ | ✓ | ✓ |
| HTTP security headers | ✓ | ✓ | ✓ |
| Email auth (SPF/DKIM/DMARC) | ✓ | ✓ | ✓ |
| Cookie security flags | ✓ | ✓ | ✓ |
| Exposed files check | ✓ | ✓ | ✓ |
| CMS version + plugin CVE (passive) | ✓ | ✓ | ✓ |
| Passive subdomain enumeration | ✓ | ✓ | ✓ |
| Active subdomain resolution | — | ✓ | ✓ |
| CDN origin IP discovery | — | ✓ | ✓ |
| Port scan + service fingerprinting | — | ✓ | ✓ |
| CVE correlation on services | — | ✓ | ✓ |
| Headless browser crawl | — | ✓ | ✓ |
| OWASP Top 10 (2025) active scan | — | ✓ | ✓ |
| WPScan / CMS deep scan | — | ✓ | ✓ |
| AI triage: duplicates merged, likely false positives flagged | ✓ | ✓ | ✓ |
| AI severity grading and plain-English report | ✓ | ✓ | ✓ |
| AI pen testing of discovered endpoints | — | — | ✓ |
| Business-logic / multi-step testing | — | — | ✓ |
| Authenticated scanning | — | — | ✓ |
| Human analyst validation (add-on, +20 credits) | — | — | ✓ |
| Free retest after remediation | — | — | ✓ |
Every Full and Premium scan is a superset of Essential — you never pay for less depth when you upgrade.
The report
Every Scantinel report is structured in three layers:
The AI: analysis on every scan, pen testing on Premium
On every scan, frontier AI models read the raw output of all scanners: they merge duplicates, flag likely false positives (marked tentative and ranked last, never deleted), and write remediation guidance. The technical facts always come from the scan data.
On Premium, an AI agent takes the full scan state (every page, API endpoint and finding) and pen tests the endpoints it found. It works in a loop: form a hypothesis, send a scoped probe, read the response, verify, report. It is looking for flaws like these:
IDOR vulnerabilities
Can user A access user B's resources by changing an ID in the request?
Multi-step auth bypass
Can a specific sequence of API calls yield an authenticated session without valid credentials?
Business logic flaws
Can the checkout flow be manipulated to apply a discount indefinitely?
Parameter pollution
Can mass assignment or parameter manipulation expose or override restricted fields?
The agent decides what to test; our own harness runs each probe after checking it against your verified scope and a non-destructive allowlist. Every step is logged. Add human validation and a security analyst confirms each finding before it reaches your report.
Questions about how it works
- Is Scantinel safe to run on a live production site?
- Yes. Scantinel uses conservative rate limits and never runs denial-of-service or destructive tests. Essential is entirely passive. Full scans generate traffic comparable to a moderately busy day of visitors.
- What is ownership verification and why is it required?
- Before any active probe runs, you prove you control the target domain via DNS TXT record, HTTP file, HTML meta tag, or email challenge. This is both a legal requirement and a core safety principle — it ensures Scantinel cannot be used against third parties.
- How does AI pen testing differ from a standard DAST scanner?
- Standard DAST scanners match requests against a fixed pattern library. The Premium AI agent reasons over the full scan state, forms a hypothesis about each discovered endpoint and sends targeted follow-up probes, verifying before it reports. That finds business-logic flaws and IDOR chains that pattern-matching tools miss.
- Does the AI run on every scan?
- Yes. Every scan, including Essential, ends with AI analysis: frontier models triage the raw results, flag likely false positives, and write remediation guidance. AI pen testing of the discovered endpoints is part of Premium, available on the Starter plan and above.
See what your website is exposing
Choose a depth and start your first scan. AI-triaged Essential results in minutes.