Run adversarial code audits with Bug Hunter
Use Bug Hunter to run scan-first adversarial code audits where Hunter, Skeptic, and Referee agents check repository bugs, security issues, and fix evidence before any approved edits.
npx skills add agentskillexchange/skills --skill run-adversarial-code-audits-with-bug-hunter
# Run adversarial code audits with Bug Hunter
What this skill actually does
Use this skill when an operator wants a structured adversarial review pass over a repository before accepting or generating fixes. Bug Hunter installs into supported coding agents and runs a pipeline where a Hunter proposes bug claims, a Skeptic challenges them, and a Referee decides what the evidence supports.
Invoke this instead of ordinary code review when the job needs a repeatable audit workflow: scanning a repository or path for security vulnerabilities, logic errors, runtime bugs, or suspicious fixes, while keeping the first run scan-only. For a reviewed fix flow, the operator can request a fix plan and explicitly approve edits.
The scope is the adversarial audit and optional approved-fix workflow. It is not a generic security scanner card, npm package listing, or autonomous editing permission grant. The default boundary is scan-only reporting with file evidence; fixes require an explicit plan and approval.
Source-backed install path from the upstream README: npx –yes https://github.com/codexstar69/bug-hunter/archive/refs/heads/main.tar.gz install –agent codex
Recommended first prompt from upstream: Use the bug-hunter skill to scan this repository. Do not edit files.
For reviewed fixes, use the documented fix-plan workflow and keep edits on a safe branch or under the host agent’s reviewed permission mode.