Analyze any repo.
Trust the result.
Sentinel Sandbox runs untrusted install scripts inside a disposable VM, monitors every system call, and delivers a plain-language safety verdict — powered by IBM Bob.
$ sentinel scan https://github.com/evil/pkg
✓ Sandbox VM created
✓ npm install captured
✓ Static analysis complete
⚠ Network call → 203.0.113.45:443
⚠ File write → /etc/cron.d/backdoor
✗ Verdict: DANGEROUS
Sandbox destroyed. Report ready.▋
7
real attack vectors detected
0
files exposed per sandbox
100%
sandbox isolated & destroyed
<2 min
average scan time
How It Works
Three steps from URL to actionable security verdict.
Submit a Repo URL
Paste any GitHub URL and choose a branch. No credentials or local setup needed.
Bob Monitors the Sandbox
Every file write, process spawn, and network call is captured live in an isolated VM.
Read the Report
Bob synthesises a plain-language verdict with full attack-chain reasoning you can act on.
Everything you need to ship safely
Disposable Sandbox
Every run gets a fresh VM destroyed the moment the scan ends.
Live Activity Monitor
Watch system calls stream in real time as the install script executes.
Bob Chain Tracing
Full attack-chain reconstructed step-by-step by IBM Bob.
Plain-Language Report
No jargon — Bob explains exactly what the code tried to do.
Interactive Q&A
Ask Bob follow-up questions grounded in the specific scan findings.
Deterministic Verdict
Safe, Suspicious, or Dangerous — always a clear, actionable answer.
Ready to scan your
next dependency?
Paste a GitHub URL and get a full AI-powered security report in under 2 minutes. No account required.