Security Audit Skill
Agent skill that runs multi-phase security audits on a codebase
What it is
A skill for coding agents that walks a repository through reconnaissance, coverage-tracked vulnerability hunting, adversarial validation by separate sub-agents, and report generation. Findings are written as confirmed, needs-validation, or rejected records in a JSON file checked against a schema, with bundled Node validators for findings and the coverage ledger. Developers and security engineers install it into their agent to audit their own code, rerunning it to fill coverage gaps across web, cloud, native, LLM, and supply-chain attack classes.