From keval-pr-review
Comprehensive PR review tailored to Klair codebase patterns. Use when reviewing a pull request, checking code quality, or running pre-merge validation. Triggers on "review PR", "review my changes", "pre-merge check", "klair review".
How this skill is triggered — by the user, by Claude, or both
Slash command
/keval-pr-review:klair-pr-reviewThis skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
Comprehensive code review skill derived from analysis of 500+ PRs and 1,988 human review comments from the Klair team. Covers both frontend (React/TypeScript) and backend (Python/FastAPI) with codebase-specific patterns.
Comprehensive code review skill derived from analysis of 500+ PRs and 1,988 human review comments from the Klair team. Covers both frontend (React/TypeScript) and backend (Python/FastAPI) with codebase-specific patterns.
1. Gather Changes ──► 2. Review (7 dimensions) ──► 3. Report
git diff main Run all applicable Structured output
Read changed files checklists in parallel by severity
# Get the diff
git diff main...HEAD --stat # Overview of changed files
git diff main...HEAD # Full diff
git log main..HEAD --oneline # Commits in this PR
Read each changed file in full (not just the diff). Many issues require surrounding context.
Run ALL 7 dimensions that apply to the changed files. Use Task agents in parallel for independent dimensions.
| # | Dimension | Applies To | Guideline |
|---|---|---|---|
| 1 | Correctness & Logic | All files | [guidelines/correctness-review.md] |
| 2 | Error Handling & Resilience | All files | [guidelines/error-handling-review.md] |
| 3 | Security | All files | [guidelines/security-review.md] |
| 4 | Frontend Patterns | .ts/.tsx/.css | [guidelines/frontend-review.md] |
| 5 | Backend Patterns | .py | [guidelines/backend-review.md] |
| 6 | Data Pipeline | pipelines/, ETL code | [guidelines/pipeline-review.md] |
| 7 | Cross-Cutting Concerns | All files | [guidelines/cross-cutting-review.md] |
Output a structured report following this format:
# PR Review: [PR title or branch name]
## Critical (must fix before merge)
- [file:line] Issue description. **Why:** Impact explanation. **Fix:** Specific recommendation.
## High (strongly recommended)
- ...
## Medium (should fix)
- ...
## Low / Suggestions
- ...
## Positive Observations
- Things done well (reinforces good patterns)
## Summary
- X critical, Y high, Z medium, W low findings
- Recommendation: APPROVE / REQUEST CHANGES / APPROVE WITH SUGGESTIONS
| Level | Criteria | Examples |
|---|---|---|
| Critical | Will cause incorrect behavior, data loss, security vulnerability, or production failure | Wrong formula, silent data corruption, XSS, unhandled crash path |
| High | Significant quality issue likely to cause problems | Silent failures, missing error propagation, React anti-patterns causing state bugs |
| Medium | Should be fixed but won't cause immediate problems | Missing validation, hardcoded values, stale comments, inconsistent patterns |
| Low | Suggestions for improvement | Naming, minor refactoring, optional type narrowing |
npx claudepluginhub ai-builder-team/ai-builder-plugin-marketplace --plugin keval-pr-reviewConducts systematic code reviews of GitHub pull requests, auditing CLAUDE.md compliance, bugs, git history, prior PR comments, and code comments for actionable feedback.
Performs thorough pull request reviews with parallel agents for bugs, security issues, guideline compliance, and error handling. Provides confidence-scored feedback and batched GitHub comments.
Reviews backend pull requests for security, performance, code quality, and testing gaps across any stack. Supports GitHub and GitLab PRs.