Target
One or more URLs — the harness recons each, then intelligently selects matching agents.
Model panel
1st = primary · others fail over & form the validator jury.
Live execution
Recon → intelligent agent selection → parallel exploitation → N-model voting → report. Artifacts saved to runs/.
— idle. Launch an engagement to stream activity. —
Validated findings
Only findings confirmed by multi-model adversarial voting appear here.
No findings yet — run an engagement.
Report
HTML report + JSON/MD artifacts for reuse by other tools/AIs.
Run an engagement to generate a report.
Agent library
…
Providers & models
Use via API key or subscription (local CLI login). CLI-capable providers are tagged.