What does 'evidence-first' reporting mean?

The skill is required to present specific code paths, scenarios, or logic flaws that prove something is wrong before it describes the associated risk.

How does it improve AI-generated code?

It specifically hunts for 'LLM slop,' which includes hallucinated APIs, context blindness, and patterns that look correct but fail in production environments.

Can I use this for non-coding tasks?

Yes, it is highly effective for architecture reviews, business proposals, and process design where you need to identify potential points of failure.

What is the core philosophy of the Frank Grimes skill?

The core assumption is that everything is 'crap' until proven otherwise. It uses a pessimistic iteration loop to move from blind optimism to earned confidence through rigorous testing.

The Grimes Grind: Hardened Code Review

Name: The Grimes Grind: Hardened Code Review
Author: misfitdev

bymisfitdev

•

Security & Testing

Systematically identifies and destroys flaws in code, architecture, and plans through a disciplined falsification process.

The Grimes Grind is a clinical, pessimistic iteration loop designed to move from blind hope to earned confidence by assuming everything is broken until proven otherwise. This skill subjects AI-generated code, architectural designs, and business proposals to a relentless destruction cycle, specifically targeting 'LLM slop,' security vulnerabilities, and reliability gaps. By forcing a confrontation with specific evidence of failure before shipping, it ensures that only the most resilient solutions survive the transition to production.

Key Features

01Severity-based issue prioritization using a unique Grimes ID system

021 GitHub stars

03Disciplined Falsification Review to prove ideas wrong before they fail in production

04Evidence-first reporting highlighting specific code paths and logic flaws

05Multi-language and multi-file analysis for complex, cross-platform projects

06Detection of 'LLM slop,' hallucinated APIs, and cargo-culted patterns

Use Cases

01Hardening security posture by identifying input validation gaps and resource leaks

02Reviewing AI-generated code to eliminate hallucinations and confident nonsense

03Conducting pre-mortems on architectural designs to identify terminal failure modes

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add misfitdev/claude-plugins frank-grimes

For use in Claude.ai and ChatGPT

Download Skill