From claude-commands
Guards against over-engineering and unnecessary code. Runs a seven-rung ladder before writing any code, every diff, fix, or feature — reuse stdlib/installed deps first, write minimum code last.
How this skill is triggered — by the user, by Claude, or both
Slash command
/claude-commands:ponytailThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
> "The best code is the code never written."
"The best code is the code never written." — Dietrich Gebert's ponytail
.github/copilot-instructions.md, adapted as a Claude/Codex skill.
Lazy means efficient, not careless. Before writing any code, stop at the first rung that holds.
Run these in order, AFTER you understand the problem — not instead of it. Read the task, trace the real flow end-to-end, then climb.
The ladder runs after you understand the problem. A small diff you don't understand is just laziness dressed up as efficiency — a second bug in a smaller package.
A report names a symptom. Find the shared function, fix it once.
This is the same discipline as ~/.claude/skills/root-cause-first/SKILL.md — patch the upstream cause, not the surface symptom.
ponytail: comment. If the shortcut has a known ceiling (global lock, O(n²) scan, naive heuristic), the comment names the ceiling and the upgrade path.You stay rigorous on:
Non-trivial logic leaves ONE runnable check behind:
Trivial one-liners need no test.
Load ponytail BEFORE:
If you find yourself reaching for a new dependency, abstraction, or framework — re-read rung 2 ("already in this codebase?") before rung 5 ("already-installed dependency?"). Order matters: in-tree reuse beats a new package.
~/.codex/skills/ponytail/SKILL.md.~/.claude/skills/root-cause-first/SKILL.md — patch the upstream cause, not the symptom. Ponytail's "fix the shared function once" rule and root-cause-first are two sides of the same coin.~/.claude/skills/zero-framework-cognition/SKILL.md — never invent heuristic/keyword/regex classification; delegate to a model. Complements ponytail's "no abstractions that weren't asked for."~/.claude/skills/code-centralization/SKILL.md — prefer one source of truth over scattered projections. Ponytail rung 2 ("reuse what's here") lives here.~/.claude/skills/code-standards/SKILL.md — dispatch adversarial review lanes (ZFC, ZFC-leveling, root-cause-first). Ponytail is the do discipline; code-standards is the check discipline.npx claudepluginhub jleechanorg/claude-commands --plugin claude-commandsEncourages minimal, YAGNI-driven solutions: stdlib over dependencies, native features over custom code, one line over many. Supports intensity levels lite, full, ultra.
Forces Claude to pause and ask if there's a cleverer, cheaper way before implementing. Helps avoid over-engineering by leveraging existing solutions, stdlib, or public APIs.
Enforces YAGNI and minimalism on demand, reducing code to the leanest correct solution. Invoke when you want to avoid over-engineering, bloat, or unnecessary dependencies.