Autonomous AI penetration testing, with proof attached.
Ryvx runs autonomous agents that recon a target, hunt vulnerabilities, and validate every finding with a working exploit — CVSS-scored, human-approved before anything fires against production, with a full audit trail after.
Explore the rest.
Recon and OSINT, vulnerability hunting and exploitation, AI/LLM security testing, reverse engineering, and reporting — verified against real code and git history, not the roadmap.
See what's shipped →Every benchmark we'd published turned out to be generated from a run that was cut short mid-investigation. They're withdrawn until we have one that isn't.
Read what happened →The desktop app and CLI stay permanently free. Starter and Pro plans add credit-based hosted scanning.
See plans →The agent graph, the PoC-or-it-didn't-happen verification gate, and the authorization challenge that runs before anything fires against a live target.
Read the docs →The Windows desktop app, the Linux CLI, and the RE sandbox guest image — free forever, no account needed for the CLI.
Get Ryvx →What's in the plan but not in the product today, listed honestly rather than implied.
See roadmap →