Loading...
Loading...
Found 7 Skills
Review a PR for code that can be deleted or simplified. Use when asked to review for simplicity or unnecessary edge-case overengineering.
LLM-as-judge methodology for comparing code implementations across repositories. Scores implementations on functionality, security, test quality, overengineering, and dead code using weighted rubrics. Used by /beagle:llm-judge command.
Apply Andrej Karpathy-inspired coding-agent guidelines in Codex. Use when writing, reviewing, debugging, or refactoring code to surface assumptions, avoid overengineering, keep edits surgical, and define verifiable success criteria.
Create critical MVP application briefs through ruthless scope-cutting dialogue. Use when user wants to turn business ideas and processes into applications. Ruthlessly cuts scope, challenges overengineering, and focuses on minimum viable product to test hypotheses. Creates structured MVP briefs in .ideas/[name]/app.md. Triggers include "build an app", "MVP for this", "what should the application do", or readiness to design application.
Review software designs, implementation plans, and implemented changes from an agile perspective to determine whether their software entropy is proportionate to current goals. Identify scope creep, overengineering, excessive defensive design, premature generalization, and hard-to-maintain mechanisms whose costs outweigh their value; provide smaller, more reversible alternatives with faster feedback. Use when the user asks for an entropy review, design simplification, complexity control, an overengineering review, or whether a mechanism is worth introducing, retaining, or expanding.
Right-sizes architecture to project scope. Prevents over-engineering by classifying projects into 6 tiers and constraining pattern choices accordingly. Use when designing architecture, selecting patterns, or when brainstorm/implement detect a project tier.
Pressure-test an existing direction before implementation.