review
系统化减少噪声并提升安全/一致性审查质量，适合安全与代码评审流程。
两阶段对抗式代码审查：先敌对枚举，再中性复审以确认或驳回结论。
做两轮的强审查：先以敌对视角把所有问题找出来，再用中立视角逐条确认或驳回结论，重点看安全、API 一致性和用户体验。默认只出报告不改代码，只有你同意才会把确认的修复做成 PR 并提交。适合想要严格、挑剔且可追溯的代码审查场景。
▸ 展开 SKILL.md 英文原文
Reports findings; by default makes no changes. A two-pass adversarial code review focused on security, API consistency/symmetry, and user experience. Pass 1 is hostile (assume the worst, surface everything). Pass 2 is a neutral audit that confirms, adjusts, or refutes each finding. Use when the user asks for a hostile, skeptical, or hard review — not a friendly pass. After the report there is an opt-in handoff to gauntlet:campaign — it can open one PR per confirmed fix (implementing the fixes) and hand them to campaign to gate and merge; decline and it stays report-only, making no source/tracked-file or GitHub changes (it may write ephemeral `.gauntlet/tmp` review scratch).
帮我安装这个 skill：https://raw.githubusercontent.com/lestrrat-ai/claude-code-plugins/main/plugins/gauntlet/skills/review/SKILL.mdcurl -fsSL "https://raw.githubusercontent.com/lestrrat-ai/claude-code-plugins/main/plugins/gauntlet/skills/review/SKILL.md"# Review Two phases, in order. The hostile pass finds; the neutral pass filters. Skipping phase 2 means delivering noise. ## Runtime Adapter This skill runs in Claude Code and Codex. - Resolve `<SKILL-PATH>` from the active `SKILL.md` path supplied by the host. Do not depend on `CLAUDE_PLUGIN_ROOT`, `PLUGIN_ROOT`, or another plugin-root environment variable. - "Subagent" means a fresh, context-isolated worker launched through the host's agent mechanism. - Dispatch independent workers in one parallel batch when the host supports it. Give each worker its scope, permissions, and output path in the prompt; do not require a host-specific agent type. - Constrain review workers to read-only