rails-code-review
Reviews Rails pull requests, focusing on controller/model conventions, migration safety, query performance, and Rails Way compliance. Covers routing, ActiveRecord, security, caching, and background jobs. Use when reviewing existing Rails code for quality, conducting a PR review, or doing a code review on Ruby on Rails (RoR) code.
How do I install this agent skill?
npx skills add https://github.com/midudev/autoskills --skill rails-code-reviewIs this agent skill safe to install?
- Gen Agent Trust Hubpass
The skill provides guidelines for reviewing Ruby on Rails code for security, performance, and conventions. It identifies standard Rails vulnerabilities like mass assignment and SQL injection. A low-severity risk of indirect prompt injection exists because the skill is designed to process untrusted code changes, though the skill itself lacks the capabilities to perform dangerous operations.
- Socketpass
No alerts
- Snykwarn
Risk: MEDIUM · 1 issue
What does this agent skill do?
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.
Pre-flight Checks
Before starting the review, verify the following:
- Rails project detected: Confirm you're in a Rails application (check for
config/application.rb,Gemfilewithrails, orbin/rails) - Diff available: Ensure you have access to the code changes (git diff, PR diff, or file paths)
- Files exist: Verify all files referenced in the diff exist in the repository
- Review scope clear: Confirm whether reviewing a full PR, a specific feature, or targeted files
If pre-flight checks fail:
- Not a Rails project → Use appropriate review skill for the technology stack
- No diff available → Request the diff or file paths from the user
- Files missing → Flag as Critical and request clarification before proceeding
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
| Area | Key Checks |
|---|---|
| Routing | RESTful, shallow nesting, named routes, constraints |
| Controllers | Skinny, strong params, before_action scoping |
| Models | Structure order, inverse_of, enum values, scopes over callbacks |
| Queries | N+1 prevention, exists? over present?, find_each for batches |
| Migrations | Reversible, indexed, foreign keys, concurrent indexes |
| Security | Strong params, parameterized queries, no html_safe abuse |
| Caching | Fragment caching, nested caching, ETags |
| Jobs | Idempotent, 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
Edge case handling:
- Empty diff: If no files changed, state "No code changes to review" and skip to conclusion
- Large diff (>50 files): Prioritize Critical checks first, then sample key files for Suggestion items; flag for targeted follow-up review
- Single file change: Apply all relevant review areas to that file; don't skip areas just because diff is small
- Test-only changes: Focus on test quality, coverage, and test organization; skip application code checks
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 escalationhtml_safeorrawapplied 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.
Output validation:
- Verify all file paths in
[file:line]references exist in the repository - Ensure line numbers are within the valid range for each file
- Check that each finding includes a required
Mitigation:field - Confirm severity sections are properly categorized
- Validate that
Actions required:section summarizes all findings accurately
If file references are invalid:
- Skip the finding and note:
[path/to/file.rb:LINE] — File not found in repository, skipping - Request clarification from the user before including in final review
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
| Skill | When to chain |
|---|---|
| rails-review-response | When the developer receives feedback and must decide what to implement |
| rails-architecture-review | When review reveals structural problems |
| rails-security-review | When review reveals security concerns |
| rails-migration-safety | When reviewing migrations on large tables |
| refactor-safely | When review suggests refactoring |
How can the creator link this skill?
Add the canonical catalog link to the repository README so users can inspect current installs and available audits. The publishing guide covers the complete discovery path.
<a href="https://skillzs.dev/skills/midudev/autoskills/rails-code-review">View rails-code-review on skillZs</a>