← ClaudeAtlas

comment-helllisted

Triggered by /comment-hell [paths]. Enforces the commenting guidelines on over-commented code — strips narrating comments so the code self-documents. Use when the user invokes /comment-hell, or says "too many comments", "clean up the comments", "comment hell". Scope: the current uncommitted diff, or the file paths passed as arguments.
orassayag/agentic-project-workflow · ★ 1 · AI & Automation · score 63
Install: claude install-skill orassayag/agentic-project-workflow
# /comment-hell — strip narrating comments The commenting guidelines already exist (global CLAUDE.md, Documentation Requirements): a comment is allowed only when it clarifies something that cannot be understood by reading the code; comments that narrate the written code are forbidden. Invoking this skill means the current change over-commented — clean it up. ## Scope The current uncommitted diff (staged + unstaged). If file paths were passed as arguments, those files in full instead. Never expand beyond that. ## How to judge each comment Apply the guidelines sentence by sentence: - Keep a sentence only if it adds a fact the code cannot express (a non-obvious *why*, a constraint, a relationship to distant code) or would be very hard to understand without it. When in doubt, delete — no comment is the right choice. - Prefer fixing the code over captioning it: rename, extract, restructure so the comment becomes unnecessary, when that's a small local change. - Variable declarations don't need describing — the name and usage carry the meaning. - Reduce a surviving comment to its irreducible *why* — one sentence; strip the narration of how the code works, even when accurate, even when it wraps a legitimate kernel. - On shared or generic code, remove implementation- or caller-specific references. - Never touch license headers, directive comments (`eslint-disable`, `@ts-expect-error`, `# noqa`, shebang lines), or doc comments a tool consumes (JSDoc consumed by the