adverse
addyosmani/adverse
Multi-agent adversarial code review skill that spawns three distinct reviewer subagents (Auditor, Adversary, Pragmatist) in parallel, conducts a cross-examination phase, and deterministically synthesizes the results.
49 8MITUpdated 2026-06-20
71
GoodOverview
Adverse orchestrates a sophisticated code review pipeline by utilizing a single underlying AI model configured with three orthogonal personas. It performs scope detection, executes independent initial reviews in parallel, runs a secondary cross-review round where personas validate or challenge each other's findings, and finally applies a deterministic Node.js synthesizer to generate ranked reports in Markdown, JSON, and HTML formats.
Capabilities
- ▸Automated scope detection (uncommitted changes, branch diffs, or full working directory)
- ▸Parallel multi-agent subagent spawning using native Claude Code agent tools
- ▸Round 2 cross-examination and finding validation/challenge round
- ▸Deterministic report synthesis into Markdown, JSON, and HTML
Best for
Conducting security-sensitive code reviews prior to merging pull requests, Multi-perspective evaluation of complex refactoring jobs touching multiple files, Gaining deeper insights into code vulnerabilities by combining auditor, adversary, and pragmatist viewpoints
Works with
Claude Code
Evoa Score breakdown= Σ (score × weight)
71Task usefulness20%77 → +15.4
実タスクにどれだけ役立つか（機能の豊富さ・用途の明確さ）。 — AIによるcapabilities/use-cases解析
Code quality15%90 → +13.5
実装・指示の品質。 — AIによるSKILL.md/README解析
Maintenance15%85 → +12.8
リポジトリがどれだけ活発に保守されているか。 — GitHub 最終push日時の新しさ
Documentation12%36 → +4.3
ドキュメントの充実度・分かりやすさ。 — README/独自要約の情報量
Security15%80 → +12.0
危険・不審な挙動が無いか。 — AIによるセキュリティレビュー
Originality10%85 → +8.5
ありふれたラッパーではない独自性。 — AIによる独自性判定
Popularity8%45 → +3.6
コミュニティの採用度。 — GitHub Stars/Forks（対数スケール）
Compatibility5%25 → +1.3
対応AIエージェントの広さ。 — AIによる対応エージェント判定
Weighted total71.3 / 100
ライセンス不明/制限あり（Red）のSkillは総合スコアに0.85倍の補正を適用します。 ランキングはこのScoreのみで決まり、広告で変わりません。 算出方法の詳細 →
Security considerations
Executes local Node helper scripts and spawns subagents to analyze local code repositories. Writes temporary review artifacts to the /tmp directory.
Categories
Summary and analysis are original content generated by AI Skills Rank. The skill's source text is not reproduced here — view it on the linked repository.