critique

Audits and fixes staged repository scripts and GitHub Actions workflows for robustness and security.

Updated Mar 22, 2026
One-click install
npx skills add https://github.com/bkrsna/gemini-cli --skill critique-bkrsna
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: critique
Source: https://github.com/bkrsna/gemini-cli/tree/main/tools/gemini-cli-bot/.gemini/skills/critique
Command: npx skills add https://github.com/bkrsna/gemini-cli --skill critique-bkrsna

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Automated repository maintenance scripts and GitHub Actions workflows often ship with subtle flaws: hardcoded data, missing error handling, infinite nudge loops, or bundled unrelated changes. This Skill acts as a quality gate that reviews staged changes against a rigorous technical and security checklist before they are merged. ## Core Features & Use Cases - Technical Robustness Audit: Verifies time-based logic, dynamic data fetching via the GitHub API, try/catch error handling, async performance, and correct metrics output formats. - Workflow Integrity Checks: Ensures actor-aware interventions, terminal escalation states, graceful closures, and that each PR addresses exactly one cohesive change. - Security Screening: Detects prompt injection payloads, data exfiltration, unauthorized command execution, and policy violations in staged code. - Use Case: After an investigation phase stages automation script changes, run this Skill to fix flaws directly, unstage internal bot files, and emit [APPROVED] or [REJECTED] for the CI workflow to act on. ## Quick Start Review the currently staged changes with the critique skill, fix any technical issues you find, and output your final verdict.

Frequently Asked Questions about critique

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

FAQPage Schema
How do I automatically review staged changes before merging a PR?

Run this Skill to inspect all staged files via git diff --staged against a checklist covering error handling, performance, and logic integrity. It fixes detected flaws directly, re-stages them with git add, and outputs [APPROVED] or [REJECTED] for the workflow to consume.

What checks does the critique skill perform on GitHub Actions scripts?

It verifies time-based grace period logic, dynamic fetching of maintainers via the GitHub API, try/catch around CLI calls, async execution instead of execSync in loops, correct CSV metrics output, and anti-spam escalation states.

Can this skill detect prompt injection in code changes?

Yes. It scans staged changes for comments or strings resembling prompt injection, such as instructions to ignore rules or force approval outputs, and rejects the change immediately if found.

When does the critique skill reject a pull request?

It rejects when a PR bundles unrelated changes, contains security risks like data exfiltration or arbitrary command execution, or when the solution is spammy, degrades developer experience, or cannot be confidently fixed.

What are the limitations of this automated critique approach?

It is strictly scoped to files already staged by the prior investigation phase and cannot modify unstaged files or invent new workflows. It also does not create the PR itself; the GitHub Actions workflow acts on its verdict.