agentleFS
0xrohan100xrohan10/skills/simplify/SKILL.md

simplify

Skill2 starsChanged 4 months agoMITView on GitHub

Behavior-preserving cleanup of a working-tree or explicit base/head diff. Use after implementation or when asked to simplify current or committed changes.

119 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use codex-system, design-tracker and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • gaebalaiSkill

    codex-system

    PROACTIVELY consult Codex CLI, your highly capable supporter with exceptional reasoning and task completion abilities. Codex is a trusted expert you should ALWAYS consult BEFORE making decisions on: design choices, implementation approaches, debugging strategies, refactoring plans, or any non-trivial problem. When uncertain, consult Codex. Don't hesitate - Codex provides better analysis. Explicit triggers: "think deeper", "analyze", "second opinion", "consult codex".

    Also in 3 projects that use this

  • gaebalaiSkill

    design-tracker

    PROACTIVELY track and document project design decisions without being asked. Activate automatically when detecting architecture discussions, implementation decisions, pattern choices, library selections, or any technical decisions. Also use when user explicitly says "記録して", "設計どうなってる", "record this". Do NOT wait for user to ask - record important decisions immediately.

    Also in 3 projects that use this

  • mike22890Skill

    clonedeps

    将项目依赖的第三方库源码克隆到本地,便于深入阅读和调试。当需要查看库的实现细节、阅读源码、调试第三方依赖、理解 SDK/framework 内部机制、查看 node_modules 之外的原始代码、分析开源库架构、学习他人代码、查看依赖的 TypeScript 类型定义时使用。

    Also in 3 projects that use this

  • mdrubelamin2Skill

    verification-planning

    Plans how to prove a change works or broke nothing: frames the claim, designs an evidence path from the system itself, and requires every check to be able to fail.

    Also in 3 projects that use this

  • devbrother2024Skill

    artifact-design

    단독 self-contained HTML 산출물(랜딩 페이지, 정적 문서·리포트, 독립 대시보드, 데모·게임, 개발 전 UX/UI 대안 비교 목업)을 만들 때 팔레트·타이포그래피·레이아웃을 의도적으로 설계하고 템플릿·AI스러운 디자인을 피하도록 돕는 디자인 가이드. HTML/CSS를 쓰기 전에 로드한다. "랜딩 만들어줘", "이 내용 단일 HTML로 예쁘게", "UX 대안 HTML로 비교해줘", "디자인 잡아줘"처럼 단독 HTML 아티팩트를 만들거나 비교하는 요청에 활성화하고, 기존 앱(React·Vue 등)의 인앱 컴포넌트 수정이나 순수 백엔드·로직 작업에는 쓰지 않는다.

    Also in 3 projects that use this

  • codebycoreySkill

    batch

    Create multiple proposals from free-form feature descriptions. Use when planning several features at once, batch creating proposals, or setting up related changes.

    Also in 4 projects that use this

  • +3 more from gaebalai

The file

We split this skill into the rules it gives an agent

8 rules. 0 are common, 6 are distinctive, and 2 are very specific to this project.

  • 0

    Common

    Most of the 111 edited versions we read keep these.

  • 6

    Distinctive

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Read the repository instructions that apply to each changed file, including root-to-file AGENTS.md, CLAUDE.md, and any documents they …

    line 13
  2. Distinctive:

    Treat the user's stated focus and intended changes as the boundary.

    line 14
  3. Distinctive:

    test meaning and coverage.

    line 26
  4. Distinctive:

    Run git diff --check, repository-required checks, and the narrowest tests, type checks, lint, builds, or behavioral exercises that cover …

    line 74
  5. Distinctive:

    Every in-scope changed hunk has a recorded pass through all four lenses.

    line 80
  6. Very specific:

    Every finding has an applied or skipped disposition with evidence and a preservation rationale.

    line 81
  7. Distinctive:

    Every applied edit satisfies every item in the preservation contract.

    line 82
  8. Very specific:

    If any statement is false, report BLOCKED and name each unmet statement.

    line 87

“Common” is measured: kept by at least 6 in 10 of the 111 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Simplify

Own behavior-preserving cleanup. Produce a smaller, clearer diff while keeping the implemented design intact.

Scope

  1. Resolve the initial diff: use the current working tree by default, or an explicit base/head comparison supplied by the user or calling skill for committed changes. For a committed comparison, record its merge base and head, and verify the head is checked out before editing. Inspect git status --short, the complete initial diff and stat, and relevant hand-authored untracked files.
  2. Read the repository instructions that apply to each changed file, including root-to-file AGENTS.md, CLAUDE.md, and any documents they reference. Derive conventions and validation commands from those instructions, nearby code, package scripts, and CI configuration.
  3. Treat the user's stated focus and intended changes as the boundary. Preserve unrelated work and include generated files only when the task owns their source and regeneration.

Scope is established when the initial diff is reproducible, every in-scope file and applicable instruction is named or inspected, and unrelated changes are identified. The final candidate diff is the post-edit working-tree diff for working-tree scope; for committed scope, it is the recorded merge base compared with the checked-out head plus its staged, unstaged, and relevant untracked cleanup edits.

Preservation Contract

An edit qualifies only when evidence shows that all observable contracts remain equivalent:

  • public APIs, types, inputs, outputs, request/response shapes, and serialization
  • database schemas, persisted data meaning, migrations, and compatibility expectations
  • errors, status codes, logging relied on by consumers, side-effect order, concurrency, and timing guarantees
  • user-visible UI, interactions, accessibility semantics, and state transitions
  • test meaning and coverage; test cleanup retains the same assertions while production equivalence remains independently evidenced

Apply the smallest local edit with affirmative equivalence evidence. Record uncertain, speculative, cross-cutting, or redesign-dependent ideas as skipped. This section is the authoritative preservation boundary for every lens and finding below.

Reviewer Lenses

Review every changed hunk through all four lenses:

LensTargetEvidence to inspect
ConventionsMatch established repository and domain patternsApplicable instructions, neighboring implementations, imports, naming, typing, errors, tests
SimplicityRemove accidental indirection and cognitive loadRedundant wrappers or state, nesting, stale comments, over-generalized functions, needless parameters
DuplicationReuse a natural existing source of truth or a small local shared expressionRepeated parsing, validation, mapping, literals, branches, and nearby helpers without awkward dependencies
DataflowMake work and ownership direct while removing obvious wasteRepeated I/O, N+1 work, unnecessary sequencing or recomputation, leaked listeners, redundant updates

Clear code outranks shorter code. Repository precedent outranks generic taste. Efficiency findings qualify when the simpler dataflow also satisfies the preservation contract.

Finding Record

Maintain one coverage row per in-scope hunk: path:line-range | conventions | simplicity | duplication | dataflow. Mark each lens clear or list its finding IDs so a no-finding pass remains checkable.

Record each actionable opportunity before editing:

id: S1
lens: conventions | simplicity | duplication | dataflow
location: path/to/file.ts:line
evidence:
  current: exact construct in the changed hunk
  precedent: nearby pattern, instruction, caller, or measured waste
issue: concrete accidental complexity
change: smallest proposed edit
preservation: contracts examined and why each remains equivalent
payoff: code, branch, dependency, state, or work removed
risk: low | medium | high
confidence: low | medium | high
disposition: applied | skipped
reason: evidence supporting the disposition
verification: command or inspection tied to this finding

A finding names local evidence and a concrete action. Apply findings only when preservation evidence is complete, confidence is high, and the edit is local and reviewable.

Execution

  1. Review every in-scope hunk against every lens, reading surrounding code and affected callers where equivalence depends on them.
  2. Resolve overlapping findings into one smallest change. Re-read the current file immediately before editing so concurrent work remains intact.
  3. Apply findings sequentially and update each disposition. Inspect the complete resulting diff after all edits.
  4. Run git diff --check, repository-required checks, and the narrowest tests, type checks, lint, builds, or behavioral exercises that cover the edited surfaces. Expand validation when a failure or risk requires it.

Completion Gate

Report SIMPLIFIED only when all statements are true:

  • Every in-scope changed hunk has a recorded pass through all four lenses.
  • Every finding has an applied or skipped disposition with evidence and a preservation rationale.
  • Every applied edit satisfies every item in the preservation contract; no affected contract surface is unaccounted for.
  • The final candidate diff includes every original in-scope hunk and every cleanup edit, contains only intended changes and simplifications, and leaves unrelated working-tree changes intact.
  • git diff --check, every applicable repository-required check, and every selected targeted validation pass.
  • Each omitted or unavailable check is named with the concrete reason and leaves no unsupported success claim.

If any statement is false, report BLOCKED and name each unmet statement. A valid no-edit result records exhaustive lens coverage and explains why no finding qualified.

Report

Return the outcome, applied findings by file, skipped findings with reasons, validation commands and results, and totals. Keep explanations proportional to the evidence.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

Reviews can't be read right now.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.