review

Analyze pull request diffs for SQL safety, race conditions, and LLM trust boundary violations.

Updated Mar 29, 2026
One-click install
npx skills add https://github.com/ibalasite/gstack --skill review-ibalasite
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: review
Source: https://github.com/ibalasite/gstack/tree/main/review
Command: npx skills add https://github.com/ibalasite/gstack --skill review-ibalasite

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Catches high-risk issues before merging by analyzing diffs against the base branch for SQL safety, race conditions, LLM output trust boundary violations, conditional side effects, scope drift, and other structural defects that tests often miss.

Core Features & Use Cases

  • Critical checks: Detects SQL injection patterns, unsafe DB writes, TOCTOU races, and missing atomic transitions.
  • Safety & LLM guards: Flags LLM-generated data being written without validation, SSRF risks, and prompt/tool boundary violations.
  • Workflow integration: Runs a checklist-driven two-pass review, integrates optional Greptile triage, can auto-fix mechanical issues and prompt the user for risky decisions. Use when preparing a PR for merge, validating migrations, or ensuring no scope creep slipped into a branch.

Quick Start

Run the review skill to analyze the current branch diff against the detected base branch and report critical and informational findings.

Frequently Asked Questions about review

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

FAQPage Schema
How do I automate PR review to catch SQL injection and race conditions before merging?

Automated PR review analyzes git diffs against the base branch to detect SQL injection, TOCTOU race conditions, and missing atomic transitions before code lands. It runs a checklist-driven two-pass review to identify structural defects that tests often miss.

What is scope drift detection in pre-landing code review?

Scope drift detection in pre-landing code review identifies unrelated changes that slipped into a branch by analyzing pull request diffs against the detected base branch. It ensures branches remain focused on their intended feature, migration, or CI changes before merging.

Can I use this automated review tool with GitHub and GitLab pull requests?

Yes, automated review works with both GitHub and GitLab pull requests by using optional gh and glab CLI tools to gather PR metadata. It analyzes local branch diffs and remote PR diffs to validate feature, migration, frontend, or backend changes.

How do I validate LLM trust boundaries and prevent unsafe data writes in pull requests?

Validating LLM trust boundaries in pull requests involves flagging LLM-generated data written without validation, detecting SSRF risks, and identifying prompt or tool boundary violations. The review process checks diffs for these safety guards alongside standard code defects.

Does pre-landing code review integrate with Greptile for triage?

Yes, pre-landing code review integrates optional Greptile triage to assist with analyzing pull request diffs. It can auto-fix mechanical issues and prompt the user with AskUserQuestion prompts for risky decisions during the two-pass review workflow.

What do I need to run a checklist-driven review on my current branch?

To run a checklist-driven review, you need access to git diffs against the detected base branch and optional gh or glab CLI for PR metadata. The review executes a two-pass analysis to report critical and informational findings on local branches or GitHub/GitLab PRs.