Enforce conservative coding practices and surgical edits for LLM-generated changes.
npx skills add https://github.com/ToXMon/tolu --skill karpathy-guidelines-toxmon
Please help me install this Agent Skill. Skill: karpathy-guidelines Source: https://github.com/ToXMon/tolu/tree/main/skills/karpathy-guidelines Command: npx skills add https://github.com/ToXMon/tolu --skill karpathy-guidelines-toxmon
These guidelines reduce common LLM-generated coding mistakes by forcing explicit assumptions, discouraging premature complexity, and requiring verifiable outcomes so AI-assisted edits are minimal, correct, and reviewable.
Apply these rules to the current coding task: state assumptions, suggest the simplest change, modify only required lines, and define one or two verifiable success criteria to confirm the result.
High-intent search queries and answers about installing and using this skill.
Apply conservative coding practices to prevent LLM code generation mistakes during refactoring: force explicit assumptions, mandate minimal-scope surgical edits, and define verifiable success criteria for reproducible test-driven changes.
Review AI-generated pull requests by checking for simplicity-first implementations, verifying only required lines were modified to prevent drive-by refactors, and confirming explicitly defined verifiable success criteria.
Enforce surgical code edits during AI feature implementation by converting requests into verifiable success criteria, preferring the minimal implementation that meets requirements, and modifying only the lines required by the task.
AI-assisted coding introduces unintended complexity by ignoring premature simplification. Limit it by applying simplicity-first guidelines that surface assumptions, present alternative interpretations, and stop to ask clarifying questions when requirements are ambiguous.
Applying conservative coding guidelines works for human-proposed changes in bug fixes by ensuring minimal-scope edits, explicit assumptions, and test-first steps so fixes can be validated independently and reproducibly.
schirrmacherGuide engineers to surface assumptions and define verifiable success criteria during code tasks.
saadjsCoordinate a test-first bug-fixing workflow with subagents to propose and verify fixes.
BrunoAMSilvaEnforce explicit assumptions and minimal edits in LLM-generated code.
marcusagmRun linting, type checking, and security audits across Node.js/TypeScript and Python projects.
