AgentHub

coding-guidelines

Anti-LLM-mistake guardrails (think before coding, simplicity first, surgical changes, goal-driven execution, fail fast …

Live
Open / InstallLast updated July 27, 2026

Description

Anti-LLM-mistake guardrails (think before coding, simplicity first, surgical changes, goal-driven execution, fail fast at trust boundaries, deterministic before probabilistic, AND structural design decisions — whether to merge/unify/rewrite/split a unit, compose-don't-merge, a prior 'killed' verdict is a hypothesis, reversibility-weighted choices). Use BEFORE writing or editing any code — feature, bugfix, refactor — during code review, AND when committing to a structural direction — "should we merge X", "should we unify Y", "rewrite vs refactor", "evaluate this approach". Triggers on "coding principles", "guardrails", "anti-LLM", "architecture decision", "should we merge", or whenever you are about to implement. NOT for routine implementation choices (which name, which algorithm, whether to add a flag) and NOT for general technical questions. Apply them on every code edit even when not named. A reusable SKILL.md agent skill by ui-HookeyChiang.

Author

ui-HookeyChiang

Platform

cli

Pricing model

free

Categories

Skills

Tags

skill
claude-skill
github
en

Capabilities

  • SKILL.md package