Lean Code

Lean Code

Aggressive YAGNI mode, on demand. The always-on baseline is rules/quality-code.md; this skill is the intensifier you invoke when minimalism matters most. The win is measured ("smallest solution that FULLY satisfies the task"), not a slogan.

The ladder — stop at the first rung that holds

1. Does this need to exist at all? Speculative need → skip it, say so in one line. (YAGNI)
2. Stdlib does it? Use it.
3. Native platform feature covers it? (`` over a picker lib, CSS over JS, a DB constraint over app code.)
4. An already-installed dependency solves it? Use it. Never add a new dependency for what a few lines do.
5. Can it be one line? Make it one line.
6. Only then — the minimum code that works.

The ladder is a reflex, not a research project: two rungs hold → take the higher one and move on.

Intensity

• lite — build what's asked, but name the leaner alternative in one line; user picks.
• full — the ladder enforced; stdlib/native first; shortest working diff. (Default.)
• ultra — YAGNI extremist: ship the one-liner and challenge the rest of the requirement in the same breath.

When NOT to be lean (load-bearing — never simplify these away)

Input validation at trust boundaries · error handling that prevents data loss · security · accessibility · the calibration real hardware needs · anything the user explicitly asked to keep. Lean ≠ incomplete: non-trivial logic leaves ONE runnable check behind (assert-demo or one small test). User insists on the full version → build it, no re-arguing.

Hard boundary — lean targets OVER-building, never completeness

This is orthogonal to thoroughness. "Minimal" means less code per branch, NOT fewer branches done or cutting corners with a hack. It must never be an excuse to under-deliver, skip verification, or leave a required branch — see rules/finish-the-task.md (its completeness + quality pillars win) and rules/quality-code.md (the no-monkey-patch pole). Mark a deliberate shortcut with a simplification: comment naming the ceiling + upgrade path.

Output

Code first. Then at most three short lines: what was skipped, when to add it. If the explanation is longer than the code, delete the explanation. Read the surrounding code / requirements before pruning — prune to the task, not to a blanket rule.

Why this works (evidence, not marketing)

AI agents measurably over-produce code (GitClear's 211M-line study: duplication up, refactoring down; Faros: +98% PRs, +91% review time, no velocity gain) — a "do I even need this?" constraint addresses a real, measured problem. But always-on minimalism is non-monotonic (arXiv 2601.22025, "When better prompts hurt": a blanket conciseness wrapper can drop accuracy/completeness), which is exactly why this is an on-demand skill + an advisory hook, never a hard gate. Success metric = rewrite/churn rate, not raw lines removed.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: AnastasiyaW
• Source: AnastasiyaW/claude-code-config
• License: MIT

Install and usage instructions live in the source repository linked above.