“Is there a vulnerability in the code we wrote?”

• Taint tracing across files and functions
• FP % score with the reasoning shown
• Pass / fail gate per project

“Is there a known vulnerability in code we imported?”

• Derived from SAST — no extra scan
• GHSA and CVE advisories with the fixed versions they publish
• Provenance back to the exact scan and finding

“Can it be exploited while it's running?”

• Baseline, quick-active and full-active depths
• Authenticated scanning via Interactive login
• Request / response evidence per finding

“What is actually vulnerable under real traffic?”

• In-process agent, JVM first
• Exact request, code path and stack trace
• Very low false positives

“What would a real attacker actually be able to do here?”

• Recon → discovery → analysis → exploitation
• Live reasoning and tool-call stream
• Proof attached to every finding