Skip to main content
plugin CX Codex

river-review — Claude Code plugin: what it does, how to install

river-review runs perspective-based code reviews in Claude Code and Codex using versioned, repo-owned skills to verify diffs against team standards.

MIT ★ 1,596 updated 2026-10-29

Install into Claude Code · Plugin marketplace All methods and platforms ↓
/plugin marketplace add hashgraph-online/awesome-codex-plugins
/plugin install river-review@awesome-codex-plugins
Open on GitHub

What river-review does

The plugin organizes review rules into versioned skills stored directly in your repository. When reviewing changes, its agent inspects the diff through distinct roles such as security, architecture, performance, testing, frontend, and adversarial review.

No dedicated LLM API key is required for normal plugin operations in Claude Code. An API key from a configured LLM provider is only necessary if you run headless jobs via GitHub Actions or the standalone river run runner.

River Review does not replace linters or static type checks. It evaluates diffs against plans and test evidence, leaving the final approval decision to humans without performing automatic merges.

What it helps with

  • audit diffs for security vulnerabilities and performance bottlenecks
  • run an adversarial review on pending code changes
  • maintain team review standards as versioned in-repo skills
  • cross-check implementation against architecture plans and test evidence

Example prompts for the agent

  • /river-review:review-local
  • /river-review:challenge
  • /river-review:review-team

How to install river-review

Plugin marketplace

/plugin marketplace add hashgraph-online/awesome-codex-plugins
/plugin install river-review@awesome-codex-plugins

Step by step, with commands for each platform: Claude Code plugins: marketplaces, installing, and building your own

Questions about river-review

How do I install river-review in Claude Code?

Add the marketplace registry with /plugin marketplace add hashgraph-online/awesome-codex-plugins, then run /plugin install river-review@awesome-codex-plugins.

What does river-review do?

It inspects code changes across multiple perspectives, including security, performance, architecture, and tests, verifying each finding against the repository diff.

Does river-review automatically merge pull requests?

No. Findings and verdicts serve as evaluation data for developers, while final merge and approval decisions remain entirely with humans.

Source text

river-review · MIT
# River Review

**レビューを、組織の判断資産へ。**  
Review Judgment as Code for AI-assisted development.

[![Documentation](https://img.shields.io/badge/docs-available-blue)](https://river-review.the3396.com/)
[![OpenSSF Best Practices](https://www.bestpractices.dev/projects/13339/badge)](https://www.bestpractices.dev/projects/13339)
[![OpenSSF Scorecard](https://api.securityscorecards.dev/projects/github.com/s977043/river-review/badge)](https://securityscorecards.dev/viewer/?uri=github.com/s977043/river-review)

![River Review logo](assets/logo/river-review-logo.svg)

日本語が正本です。[English README](./README.en.md)

River Review は、レビュー基準を **versioned / repo-owned な Skill** として管理し、計画・差分・テスト・既存レビューの根拠を横断して実行する OSS です。Claude Code / Codex のローカルレビューと GitHub Actions に、同じチームの判断基準を持ち込めます。人間の承認や自動マージの代行はしません。

- **Skills define judgment**—チームがレビュー観点を所有し、fixture / golden output で評価する。
- **Gates execute judgment**—計画・差分・検証証跡を対象に、現在利用可能なレビューゲートで判断基準を適用する。
- **Riverbed remembers judgment**—決定とフィードバックを記録し、不要な指摘の再発を抑える。

[設計思想](https://river-review.the3396.com/explanation/concept/) / [機能と実行モデル](https://river-review.the3396.com/explanation/what-is-river-review/) / [既知の制限](https://river-review.the3396.com/reference/known-limitations/)

## はじめる

### Claude Code(最短)

```text
/plugin marketplace add s977043/river-review
/plugin install river-review@river-review-marketplace
/reload-plugins
/river-review:review-local
```

### Codex

```text
codex plugin marketplace add s977043/river-review
```

通常のプラグイン利用では、**River Review 専用の LLM API キーは不要**です。GitHub Actions や standalone `river run` のヘッドレス実行では、設定した LLM プロバイダーの API キーが必要です。

導入手順の詳細は [プラグインを導入する](https://river-review.the3396.com/guides/plugin/install/)、GitHub Actions 設定は [5分クイックスタート](https://river-review.the3396.com/guides/quickstart/) を参照してください。npm パッケージの公開は配布方針に含まれず、リポジトリ内の開発用 CLI は `npm run river -- ...` で実行します。

<!-- x-release-please-start-version -->

最新リリース: [v1.137.0](https://github.com/s977043/river-review/releases/latest)

<!-- x-release-please-end -->

## 得られるもの

得られるもの(プラグイン名で名前空間化されます):

- コマンド: `/river-review:setup-team` / `/river-review:review-local` / `/river-review:review-team` / `/river-review:challenge` が含まれる。加えて `/river-review:skill` / `/river-review:check` / `/river-review:pr` も使える。

- スキル: オーケストレーターの `river-review` に加えて `river-review-code` / `river-review-security` / `river-review-security-audit` / `river-review-performance` の各スキルが含まれる。さらに `river-review-architecture` / `river-review-testing` / `river-review-frontend` も含まれる。加えて `river-review-docs` / `adversarial-review` / `review-team` / `unknown-coverage-review` も含まれる。いずれも `/river-review:<skill-name>` で呼び出せる

スキルを追加する前に、[Starter Cookbook](https://river-review.the3396.com/guides/starter-cookbook/) で既存の観点を試してください。プロジェクト固有の基準は [Skill作成チュートリアル](https://river-review.the3396.com/tutorials/creating-your-first-skill/) に沿って作成できます。

## ドキュメント

- **初回導入**:[Quickstart](https://river-review.the3396.com/guides/quickstart/)
- **レビュー運用**:[2段構えレビューゲート](https://river-review.the3396.com/guides/two-stage-review-gate/)
- **GitHub Actions**:[GitHub Actionsガイド](https://river-review.the3396.com/guides/github-actions/)
- **Skill一覧とスキーマ**:[Skill Catalog](https://river-review.the3396.com/reference/skills-catalog/) / [Schema](https://river-review.the3396.com/reference/skill-schema/)
- **CLIの仕様**:[CLIリファレンス](https://river-review.the3396.com/reference/runner-cli-reference/)
- **トラブルシューティング**:[Troubleshooting](https://river-review.the3396.com/guides/troubleshooting/)
- **ロードマップ**:[ROADMAP.md](./ROADMAP.md) / [Issues](https://github.com/s977043/river-review/issues)

公式ドキュメントの正本は `pages/` です。`docs/` は開発・保守資料であり、公式サイトには掲載しません。ただし両ディレクトリとも公開GitHubリポジトリ上で閲覧可能です。[ドキュメント管理方針](docs/policy/DOCUMENTATION.md)

## FAQ

**ESLint・型チェック・静的解析の代わりですか?** いいえ。River Review は計画と実装、テストと期待値など、複数アーティファクトを跨ぐレビュー判断を補完します。

**レビュー結果が自動承認になりますか?** いいえ。findings と verdict は判断材料であり、最終承認・GO / NO-GO は人間・呼び出し側の責務です。

その他は [FAQ](https://river-review.the3396.com/guides/faq/) と [セキュリティモデル](https://river-review.the3396.com/explanation/security-model/) を参照してください。

## コントリビューションとセキュリティ

バグ報告、改善提案、ドキュメントの貢献を歓迎します。

- [貢献ガイド](.github/CONTRIBUTING.md) / [行動規範](.github/CODE_OF_CONDUCT.md)
- [脆弱性の非公開報告](.github/SECURITY.md)(公開Issueへ脆弱性情報を投稿しない)
- [公開の実践進化方針](https://river-review.the3396.com/guides/governance/practice-evolution/)

## ライセンス

コードは [LICENSE-CODE](LICENSE-CODE)(MIT)、ドキュメントやメディアは [LICENSE-CONTENT](LICENSE-CONTENT)(CC BY 4.0)を参照してください。ルートの [LICENSE](LICENSE) 本文は MIT です。ただし設定ファイル等への適用範囲については、LICENSE-CODE / LICENSE-CONTENT の説明に Apache-2.0 への不一致があり、[Issue #2721](https://github.com/s977043/river-review/issues/2721) で確認中です。

© river-review maintainers · MIT · hashgraph-online/awesome-codex-plugins

Similar plugins

All plugins for Code review and refactoring →
ub-mode unclebob The ub-mode skill for Claude Code helps developers structure code changes. It guides the development process by following the principles of "First Make it Work, Then Make it Right, Then Follow Practice." CX Codex +1 930installs methodology-advisor FlorianBruniaux The methodology-advisor command for Claude Code and Codex reads the codebase and asks three questions to recommend an AI-development methodology stack. Run it at the start of a project, before choosing a workflow. CX Codex +1 ★ 6,529repo stars superpowers obra superpowers — a plugin for Claude Code and Codex that teaches the agent a complete methodology: requirements gathering, design, TDD, and subagent-driven development. Triggers automatically when you start building something new. CX Codex ★ 323,788repo stars ponytail DietrichGebert ponytail is a Claude Code plugin that steers coding tasks toward short, practical solutions. Its review command checks related code for bugs, security issues, and missing tests. ★ 196,420repo stars mattpocock-skills Matt Pocock mattpocock-skills — a set of skills for grilling, TDD, code review, and specification work. Use it when an agent needs to clarify the task before changing the code. ★ 329,816repo stars understand-anything Egonex-AI understand-anything turns codebases or markdown wikis into an interactive knowledge graph with a browser dashboard. Helps navigate unfamiliar repositories without reading files blindly. ★ 93,566repo stars serena official Oraios serena — a Claude Code plugin that integrates language server protocol for semantic analysis. Provides refactoring suggestions, code navigation, and symbol lookups across the codebase. ★ 47,914repo stars security-guidance official Anthropic security-guidance — a Claude Code plugin that reviews generated code in three layers. Regex warnings fire on edits, an LLM checks the diff after every turn, and an agentic reviewer walks related files before commits. ★ 47,914repo stars

From the same set391

The whole set →
simple-man The simple-man plugin cuts praise, recaps, and filler from agent replies while preserving actionable details. Use it when coding responses need less noise. CX Codex ★ 1,596repo stars sealos Deploys applications to Sealos Cloud directly from an agent, configuring databases, storage, and HTTPS with readiness checks. CX Codex ★ 1,596repo stars runtype-skills A skill bundle for building and operating AI products on Runtype, covering platform work, agent tools, Persona widgets, templates, and SDK workflows. CX Codex ★ 1,596repo stars roadmapsmith roadmapsmith keeps ROADMAP.md honest: one command writes it, the other proposes edits backed by real code and waits for your ok. CX Codex ★ 1,596repo stars reviewable-html-workbench The reviewable-html-workbench plugin turns agent-produced HTML into browser-reviewable documents with inline comments and follow-up edits. Use it for reports, design docs, and comparison tables. CX Codex ★ 1,596repo stars research-agent-skills Loads 281 agent skills for the full academic research lifecycle into Claude Code and Codex. Covers literature search, hypothesis generation, data analysis, citation checks and peer review. CX Codex ★ 1,596repo stars repopilot repopilot takes a repository snapshot at session start and blocks the agent from finishing if it weakens tests or removes security checks. CX Codex ★ 1,596repo stars rekall The rekall plugin compacts conversation history in the Codex VS Code extension, saves a written task handoff, and resumes work once. CX Codex ★ 1,596repo stars

Project sponsors

Companies that keep this analytics open