review

Analyze git diffs for SQL injection, race conditions, and LLM trust boundary violations.

Updated May 22, 2026
One-click install
npx skills add https://github.com/shekerkamma/peopletech-marketplace --skill review-shekerkamma
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: review
Source: https://github.com/shekerkamma/peopletech-marketplace/tree/main/plugins/gstack/review
Command: npx skills add https://github.com/shekerkamma/peopletech-marketplace --skill review-shekerkamma

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Pull requests often contain subtle but critical bugs like SQL injection, race conditions, and LLM trust boundary violations that manual review misses. This skill provides a systematic, two-pass review process to catch these issues before code merges, preventing production incidents and data loss.

Core Features & Use Cases

  • Two-Pass Review: First pass flags critical issues (SQL safety, race conditions, shell injection, enum completeness). Second pass catches informational issues (async/sync mixing, dead code, LLM prompt issues, type coercion).
  • Specialist Subagents: Delegates to parallel specialists for testing gaps, maintainability, performance, security, data migration, API contracts, and red team adversarial analysis.
  • Fix-First Workflow: Auto-fixes mechanical issues (N+1 queries, magic numbers, dead code) while escalating security and design decisions to the user.
  • Greptile Integration: Fetches and triages Greptile bot comments on GitHub PRs, classifying them as valid, false positives, or already fixed.

Quick Start

Ask Claude to review your current pull request diff for SQL injection risks, race conditions, LLM trust boundary violations, and other structural issues before you merge the code.

Frequently Asked Questions about review

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

FAQPage Schema
How do I review a git diff for SQL injection and race conditions before merging?

A pre-landing pull request review analyzes your git diff against the base branch to identify SQL injection risks, race conditions, and LLM trust boundary violations before code merge. This two-pass process catches critical bugs early to prevent production incidents.

What is a two-pass code review process for pull requests?

A two-pass code review process first flags critical issues like SQL safety, shell injection, and race conditions, then catches informational issues like async/sync mixing, dead code, and type coercion. This methodical approach ensures both critical and minor structural problems are caught before landing changes.

How do I catch LLM trust boundary violations in a pull request?

To catch LLM trust boundary violations in a pull request, analyze the git diff to identify where untrusted inputs cross into LLM processing logic. Pre-landing review heuristics distinguish these security findings from auto-fixable mechanical issues, escalating them for human judgment before merge.

Does this pre-landing review workflow integrate with Greptile comments on GitHub?

Yes, the pre-landing review workflow fetches and triages Greptile bot comments on GitHub pull requests, classifying each comment as a valid issue, a false positive, or already fixed. This integration streamlines automated bot feedback into your manual review process.

Can I automatically fix mechanical code issues found during a pull request review?

Yes, you can automatically fix mechanical issues like N+1 queries, magic numbers, and dead code found during a pull request review. The fix-first workflow auto-resolves these structural problems while escalating complex security and design decisions to the user.

What are the limitations of automated security detection in git diff parsing?

Automated security detection in git diff parsing cannot resolve judgment-based security findings or design decisions, requiring human escalation. It is limited to identifying known patterns like SQL injection and race conditions, meaning novel or context-dependent vulnerabilities may be missed.