AI-Assisted Deterministic Security Review

Evidence-Backed AI Security Reviews

Scan a public GitHub repository using deterministic security rules, then receive bounded, category-aware AI guidance based only on verified evidence.

  • ✓ Deterministic vulnerability detection
  • ✓ Evidence-backed AI explanations
  • ✓ Human-verifiable remediation guidance
Scan Public Repository

Authorization-focused controlled demo

A controlled BOLA and ownership-validation example. Fixture code is never executed.

Multi-rule controlled demo

Controlled secrets, JWT, CORS, redirects, command execution, filesystem, rate limiting, and file upload checks. Fixture code is never executed.

Both demos are controlled local static-analysis targets. Sentinel never installs dependencies or runs fixture code.

Scan Public Repository

Only public GitHub HTTPS repository URLs are supported.

Sentinel clones and statically analyzes the repository. It never installs dependencies or executes repository code.

Large repositories may exceed the bounded download limit.

Architecture and trust boundary

Evidence first. AI guidance second.

  1. 1Repository
  2. 2Deterministic Security Engine
  3. 3Evidence Package
  4. 4AI Security Review
  5. 5Dashboard
  • ✓ Static analysis only
  • ✓ Repository code is never executed
  • ✓ Public GitHub HTTPS repositories only
  • ✓ Secret values are redacted
  • ✓ Deterministic findings remain authoritative
  • ✓ AI guidance is advisory
  • ✓ Patches require human review

Sentinel’s security coverage is intentionally bounded and does not represent a comprehensive security assessment.