Skip to content
Learn

How Rokkhe works

Short explainers on the questions people ask before trusting an AI pentester: what it proves, how a finding is confirmed, how a fix is verified, and what stays inside your network. Each one starts with a direct answer and links to the docs for detail.

AI penetration testing vs DAST vs vulnerability scanningWhat a pattern-matching scanner flags, what an AI penetration test attempts, and the evidence a Rokkhe finding has to carry before it is reported.How Rokkhe validates a findingHypothesis, independent validator, reproduction: how a Rokkhe finding is accepted, rejected or left inconclusive, and what each decision keeps.How a retest verifies a fixHow a Rokkhe retest replays an accepted finding's proof and bypass attempts after a fix, who approves it, and what each verdict requires.Self-hosted AI pentesting: what runs where and what leaves your networkWhere each part of a self-hosted Rokkhe deployment runs, how scan sandboxes are isolated, and every outbound connection the stack makes.AI pentesting on every pull requestRun Rokkhe on pull requests from your own CI job with diff scope, a spend ceiling and exit codes, or let the GitHub App review each pull request inline.