Token导航 LogoToken导航TokenDH.com
研究检索只读github未标认证来源可访问许可证需确认审计通过

rails-code-reviewRails 代码审查

Agent Skill

rails-code-review 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

总安装

894

周安装

38

GitHub Stars

16

下载量

313
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

unknown

最后核验

2026-05-01

来源状态

来源可访问

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:rails-code-review(Rails 代码审查)
来源仓库:https://github.com/igmarin/rails-agent-skills
仓库路径:skills/rails-code-review
安装命令:
npx skills add https://github.com/igmarin/rails-agent-skills --skill rails-code-review
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/igmarin/rails-agent-skills --skill rails-code-review

简介

用于辅助 Rails 代码审查,识别潜在缺陷与风格不一致问题。

  • 适合在 Pull Request 合并前自动化检查常见错误模式。
  • 通过 GitHub 安装,支持对比分支差异与规则集匹配。
  • 不能完全替代人工评审,尤其涉及业务逻辑与安全相关代码。
  • 建议集成静态分析工具,补充语法与复杂度检测。rails-code-review 属于研究检索类 Skill,可作为该场景下的辅助能力补充。

SKILL.md

Rails Code Review (The Rails Way)

When reviewing Rails code, analyze it against the following areas. When writing new code, follow rails-code-conventions (principles, logging, path rules) and rails-stack-conventions (stack-specific UI and Rails patterns).

Core principle: Review early, review often. Self-review before PR. Re-review after significant changes.

HARD-GATE: After implementation (before PR)

After green tests + linters pass + YARD + doc updates:
1. Self-review the full branch diff using the Review Order below.
2. Fix Critical items; resolve or ticket Suggestion items.
3. Only then open the PR.
generate-tasks must include a "Code review before merge" task.

Quick Reference

AreaKey Checks
RoutingRESTful, shallow nesting, named routes, constraints
ControllersSkinny, strong params, before_action scoping
ModelsStructure order, inverse_of, enum values, scopes over callbacks
QueriesN+1 prevention, exists? over present?, find_each for batches
MigrationsReversible, indexed, foreign keys, concurrent indexes
SecurityStrong params, parameterized queries, no html_safe abuse
CachingFragment caching, nested caching, ETags
JobsIdempotent, retriable, appropriate backend

Review Order

Work through the diff in this sequence. Deep criteria: REVIEW_CHECKLIST.md. One-page PR baseline: assets/checklist.md. Finding examples (JSON + comment shape): assets/examples.md.

Configuration → Routing → Controllers → Views → Models → Associations → Queries → Migrations → Validations → I18n → Sessions → Security → Caching → Jobs → Tests

Critical checks to spot immediately:

# N+1 — one query per record in a collection
posts.each { |post| post.author.name }       # Bad
posts.includes(:author).each { |post| post.author.name }  # Good

# Privilege escalation via permit!
params.require(:user).permit!                # Bad — never in production
params.require(:user).permit(:name, :email)  # Good

Always Critical (flag every occurrence as Critical):

  • params.require(...).permit! — mass-assignment / privilege escalation
  • html_safe or raw applied to user-supplied content — XSS
  • Missing authorization check on a sensitive action
  • Business logic inside a controller action — pricing, tax, discount, multi-step workflow, or any domain calculation inline. A controller action that does more than coordinate (call one service, render response) is Critical, not a Suggestion.
  • Unparameterized / string-interpolated SQL — injection
  • Destructive migration without a safe path on large tables

Severity levels

Use only these labels (no High/Low, P0–P2, etc.): Critical | Suggestion | Nice to have.

  • Critical — security, data loss, crash, or any Always Critical rule → block merge; re-diff after fix.
  • Suggestion — conventions / performance → fix in PR, or ticket if redesign is large.
  • Nice to have — small style or micro-optimization → optional for the author.

Output style

Group findings under ### Critical / ### Suggestion / ### Nice to have (omit empty sections). Do not use a single flat list mixed by severity.

## Review — <PR title or area>

### Critical
- [path/to/file.rb:LINE] (Area) One-line risk. **Mitigation:** concrete next step.

### Suggestion
- [path/to/file.rb:LINE] (Area) … **Mitigation:** …

### Nice to have
- …

**Actions required:** <one line per severity level that appeared — e.g. Critical → block merge + re-review; Suggestion → …>

Template rules: each bullet is [file:line] (Area) + risk + Mitigation: (required). Tag (Area) from: Controllers, Routing, Views, Models, Queries, Migrations, Validations, Security, Caching, Jobs, Tests — across the whole review, cover ≥4 distinct areas when the diff touches that many surfaces.

Re-review before merge

Re-diff the branch after any Critical fix (mandatory), after >3 Suggestion fixes or any logic/architecture change during feedback (recommended), or whenever the fix could alter queries, auth, or migrations. Skip only for Nice to have-only feedback or trivial one-line edits with no behavior change.

Review anti-patterns (adds to checklist, does not replace it)

  • Thin controller → fat model: extract orchestration to services (PORO / *.call), not giant model methods.
  • N+1 in dev: small seeds hide N+1 — if associations run inside a loop, count queries (request spec, rack-mini-profiler, logs) instead of assuming “it’s fast here.”
  • Hot-table migrations: add concurrent indexes and heavy backfills in separate deploy steps from reversible schema changes (chain rails-migration-safety when unsure).
  • Callbacks vs jobs: persistence hooks only; external I/O and multi-step workflows belong in services/jobs with clear idempotency.

Integration

SkillWhen to chain
rails-review-responseWhen the developer receives feedback and must decide what to implement
rails-architecture-reviewWhen review reveals structural problems
rails-security-reviewWhen review reveals security concerns
rails-migration-safetyWhen reviewing migrations on large tables
refactor-safelyWhen review suggests refactoring

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Codex

33.55%
按下载量换算105

Claude

31.98%
按下载量换算100

Cursor

18.12%
按下载量换算57

Gemini CLI

9.46%
按下载量换算30

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

只读

该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills