agentleFS
Sign inStart free
omspradippatilomspradippatil/Claude-Skills/.agents/skills/ponytail/SKILL.md

ponytail

Skill3 starsChanged 10 days agoMITView on GitHub

Activate this skill for minimal engineering, code simplification, refactoring to reduce complexity, eliminating unnecessary abstractions, wrappers, or dead dependencies, and enforcing the "lazy senior developer" philosophy (YAGNI, standard library and native platform preference).

We split this skill into the rules it gives an agent

15 rules. 0 are common, 9 are distinctive, and 6 are very specific to this project.

  • 0

    Common

    Most of the 130 edited versions we read keep these.

  • 9

    Distinctive

    This author's own choices. Worth a look.

  • 6

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Before writing any new code, evaluate solutions in this strict order:

    line 35
  2. Distinctive:

    No wrapper components without clear value: Avoid creating single-use wrappers around standard components or libraries.

    line 57
  3. Distinctive:

    No multi-layered architectures for simple tasks: Avoid creating a repository, service, interface, DTO, and controller for a simple 10-line …

    line 58
  4. Distinctive:

    Flatten unnecessary indirection: If a helper function is only called once and adds no semantic clarity, inline it.

    line 59
  5. Distinctive:

    Avoid speculative configuration: Do not add dynamic config switches, feature flags, or factory patterns for variations that do not …

    line 60
  6. Very specific:

    Frontend: Prefer HTML5 semantic elements (<dialog>, <details>, <input type="date">), CSS Grid/Flexbox, native CSS variables, and modern Web …

    line 63
  7. Very specific:

    Node.js: Prefer built-in node:fs/promises, node:path, node:crypto, node:test, fetch over external packages like lodash, rimraf, uuid, axios …

    line 64
  8. Very specific:

    Python: Prefer pathlib, dataclasses, functools, typing, json, sqlite3 over redundant external packages.

    line 65
  9. Distinctive:

    If the project already uses a specific validation library (e.g., Zod), do not introduce a second one (e.g., Joi or Yup).

    line 69
  10. Distinctive:

    Retain full input validation and sanitization: Never bypass security controls to reduce lines of code.

    line 74
  11. Distinctive:

    Preserve accessibility (a11y): Never remove ARIA attributes, semantic tags, or focus outlines.

    line 77
  12. Very specific:

    Ponytail approach: Use native Intl.DateTimeFormat or standard library methods.

    line 86
  13. Very specific:

    Ponytail approach: Use native structuredClone() (standard in modern browsers and Node.js 17+).

    line 90
  14. Very specific:

    Ponytail approach: Use standard HTML5 <dialog> element with .showModal() and native CSS backdrop styling.

    line 94
  15. Distinctive:

    When you make an intentional design trade-off (e.g., inlining code or choosing stdlib over a library), leave a brief 1-line code comment or …

    line 100

“Common” is measured: kept by at least 6 in 10 of the 130 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

Ponytail — Minimal Engineering Skill

Official upstream: DietrichGebert/ponytail (MIT License)

Philosophy & Core Objective

The best code is the code you never wrote. Every line of added code is a liability to be read, tested, debugged, and maintained.

The goal of Ponytail is to enforce disciplined minimalism: eliminate premature abstractions, avoid speculative complexity, reject unnecessary third-party dependencies, and prefer native platform capabilities.

When to Activate

  • Refactoring code to simplify architecture and eliminate bloat.
  • Reviewing PRs or proposed designs for over-engineering.
  • Implementing features where minimal footprint and maximum maintainability are priorities.
  • Evaluating whether to add a new third-party dependency versus using existing libraries or native APIs.
  • Removing dead code, redundant wrappers, and speculative configuration layers.

The Decision Ladder

Before writing any new code, evaluate solutions in this strict order:

1. YAGNI (Does this problem need code at all?)
   └── If no: Stop. Do not build it.
2. Standard Library (Can Python / Node.js / Go / Rust stdlib solve it?)
   └── If yes: Use stdlib. Add zero dependencies.
3. Native Platform Feature (HTML5, modern CSS, Web APIs, shell builtins)
   └── If yes: Use the platform primitive directly.
4. Already-Installed Dependency (Can existing packages in package.json/pyproject.toml do it?)
   └── If yes: Reuse the established tool.
5. In-Place Idiomatic Code (Can it be solved in 5-15 readable lines?)
   └── If yes: Implement directly without creating a new class/module.
6. Minimum Viable Code (Write only what satisfies the acceptance criteria)
   └── Implement cleanly with no speculative generality.

Implementation Rules

1. Reject Premature Abstractions

  • No wrapper components without clear value: Avoid creating single-use wrappers around standard components or libraries.
  • No multi-layered architectures for simple tasks: Avoid creating a repository, service, interface, DTO, and controller for a simple 10-line database query.
  • Flatten unnecessary indirection: If a helper function is only called once and adds no semantic clarity, inline it.
  • Avoid speculative configuration: Do not add dynamic config switches, feature flags, or factory patterns for variations that do not currently exist.

2. Native-First Choices

  • Frontend: Prefer HTML5 semantic elements (<dialog>, <details>, <input type="date">), CSS Grid/Flexbox, native CSS variables, and modern Web APIs (fetch, Intl, URLSearchParams) over heavy utility packages.
  • Node.js: Prefer built-in node:fs/promises, node:path, node:crypto, node:test, fetch over external packages like lodash, rimraf, uuid, axios unless project requirements dictate otherwise.
  • Python: Prefer pathlib, dataclasses, functools, typing, json, sqlite3 over redundant external packages.

3. Reuse Established Project Patterns

  • Inspect existing codebase conventions first.
  • If the project already uses a specific validation library (e.g., Zod), do not introduce a second one (e.g., Joi or Yup).
  • Match the established project architecture rather than inventing a competing paradigm.

4. Non-Negotiable Safeguards

Minimal code does not mean cutting corners on safety:

  • Retain full input validation and sanitization: Never bypass security controls to reduce lines of code.
  • Retain necessary error handling: Handle known error modes gracefully.
  • Preserve type safety and tests: Maintain complete type annotations and unit test coverage.
  • Preserve accessibility (a11y): Never remove ARIA attributes, semantic tags, or focus outlines.
  • Do not write unreadable one-liners: Clean, readable code over clever, unmaintainable code golf.

Practical Examples

Example 1: Date & Time Manipulation

  • Over-engineered: Installing moment.js (large legacy bundle) just to format a date string.
  • Ponytail approach: Use native Intl.DateTimeFormat or standard library methods.

Example 2: Deep Clone

  • Over-engineered: Importing lodash.clonedeep for JSON-serializable state cloning.
  • Ponytail approach: Use native structuredClone() (standard in modern browsers and Node.js 17+).

Example 3: Modal Dialog

  • Over-engineered: Pulling in a 20KB third-party modal library with 10 props and portal wrappers.
  • Ponytail approach: Use standard HTML5 <dialog> element with .showModal() and native CSS backdrop styling.

Architectural Rationale

When you make an intentional design trade-off (e.g., inlining code or choosing stdlib over a library), leave a brief 1-line code comment or report summary so future developers understand the reasoning.

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.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

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.

Check your setup, free