What problem does it solve? Manual security reviews miss subtle bugs and automated scanners only catch known patterns. This Skill orchestrates a structured, multi-phase security audit that finds exploitable vulnerabilities with real impact — including business logic flaws, chained attacks, and trust-boundary violations that scanners cannot detect. ## Core Features & Use Cases - Six-phase audit workflow: reconnaissance, parallel vulnerability hunting, adversarial validation, reporting, schema-validated structured output, and independent verification of every finding. - Domain-specific attack classes: dedicated hunting guides for web apps, memory-safety/binary/kernel targets, AI/LLM agents, HTTP protocol and auth (JWT, OAuth, SAML), and client-side browser code. - Machine-readable results: produces REPORT.md, FINDINGS-DETAIL.md, and a schema-validated findings.json with traces, severities, and remediations. - Use Case: Point it at a web API repository before a release; it maps trust boundaries, launches parallel hunting agents across injection, access control, and business logic classes, then returns a report containing only confirmed, exploitable findings with concrete attack scenarios. ## Quick Start Run a security audit on the codebase in the current directory and write the report to a new output folder.
