review

Analyze PR diffs for SQL safety, LLM trust boundaries, and side effects.

Updated Feb 27, 2026
One-click install
npx skills add https://github.com/abustamam/portfolio-blog --skill review-abustamam
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: review
Source: https://github.com/abustamam/portfolio-blog/tree/main/.claude/skills/gstack/review
Command: npx skills add https://github.com/abustamam/portfolio-blog --skill review-abustamam

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

The review skill analyzes diffs against main during PRs to surface SQL safety risks, LLM trust boundary issues, and potential side effects that could degrade correctness or security.

Core Features & Use Cases

  • Diff-based PR analysis to detect SQL safety issues, data integrity risks, and unsafe patterns.
  • LLM trust boundary checks and guidance for mitigating prompt-injection risks in review comments.
  • Structured triage workflow (Greptile-style) to classify findings and guide reviewer actions.

Quick Start

Run the /review workflow on your current branch to get a structured risk assessment of the PR diff.

Frequently Asked Questions about review

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I detect SQL safety risks in a pull request diff?

You can detect SQL safety risks in a pull request diff by running the review workflow, which analyzes changes against main to surface data integrity issues and unsafe SQL patterns for pre-landing guidance.

What is LLM trust boundary analysis for prompt-injection risks?

LLM trust boundary analysis checks PR diffs to identify prompt-injection risks, guiding reviewers on mitigating security issues where untrusted inputs might compromise language model behavior.

How do I check PR diffs for conditional side effects before landing code?

To check PR diffs for conditional side effects, run the review workflow to analyze state-transition issues and flag potential correctness or security degradations before merging.

Can I use Greptile-style triage to categorize code review findings?

Yes, the review workflow uses Greptile-style triage to classify findings from PR diff analysis, categorizing SQL safety, LLM trust boundary, and side effect issues to guide reviewer decisions.

Does the review workflow require any external dependencies to analyze PRs?

No, the review workflow requires no external dependencies, operating as a standalone skill to analyze branch diffs against main for SQL, LLM trust, and side effect risks.

What is the best way to automate risk assessment for pre-landing code reviews?

The best way to automate pre-landing risk assessment is running the review workflow on your current branch, which provides a structured triage of SQL, LLM, and state-transition findings.