agentleFS
S3n0001S3n0001/agent-skills/ponytail/SKILL.md

ponytail

Skill1 starsChanged 2 months agoMITView on GitHub

Apply opt-in minimalism when the user invokes $ponytail or asks for Ponytail or lazy mode. Prefer reuse, deletion, platform features, and the smallest correct diff. Never auto-trigger for ordinary work.

158 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use reimagine, herdr-cli 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.
  • dev-willbird1936Skill

    reimagine

    Rebuild an instruction file for reliable adherence and lower context cost. Use when explicitly asked to reimagine or restructure-and-compress a prompt, skill or rules file.

    Also in 3 projects that use this

  • julweberSkill

    herdr-cli

    Expert guide for herdr, the terminal workspace manager for AI coding agents.

    Also in 3 projects that use this

  • vikas8520-coderSkill

    cli-backup-sync

    Backup and restore Nerve data from the CLI. Trigger incremental snapshots, sync to cloud storage, manage backup schedules, and restore from archive files.

    Also in 3 projects that use this

  • vikas8520-coderSkill

    cli-batches

    Submit and monitor batch inference jobs from the CLI. Upload and manage files for batch processing, retrieve results, and integrate batch pipelines with CI/CD workflows.

    Also in 3 projects that use this

  • +8 more from vikas8520-coder

The file

We split this skill into the rules it gives an agent

12 rules. 0 are common, 12 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 153 edited versions we read keep these.

  • 12

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Stop at the first rung that fully satisfies the request:

    line 12
  2. Distinctive:

    Does the codebase already have it?: Search and reuse the owning helper, type, pattern, or seam.

    line 15
  3. Distinctive:

    Does the standard library solve it?: Use it.

    line 16
  4. Distinctive:

    Does a native platform feature solve it?: Prefer HTML, CSS, database, framework, or operating-system behavior over custom code.

    line 17
  5. Distinctive:

    Can one readable local expression solve it?: Keep it local.

    line 19
  6. Distinctive:

    Only then: write the minimum new code that is correct.

    line 20
  7. Distinctive:

    Prefer deletion over addition and boring code over clever code.

    line 28
  8. Distinctive:

    Avoid one-implementation interfaces, one-product factories, speculative configuration, parallel frameworks, scaffolding "for later," and …

    line 29
  9. Distinctive:

    Use the fewest files that produce one coherent, readable outcome.

    line 30
  10. Distinctive:

    When two small solutions are equivalent, prefer the one with safer edge behavior and lower ownership cost.

    line 31
  11. Distinctive:

    Add more checks only for materially distinct risks or acceptance conditions.

    line 41
  12. Distinctive:

    Do not introduce a test framework for one change.

    line 42

“Common” is measured: kept by at least 6 in 10 of the 153 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

Apply an opt-in minimalism mode to coding. Lazy means efficient, not careless: understand the real flow first, then minimize the solution. Default to full intensity for the current task only. Honor lite, full, or ultra when named, and stop when the user asks for normal mode.

Climb the ladder

Stop at the first rung that fully satisfies the request:

  1. Does it need to exist? Skip speculative work.
  2. Does the codebase already have it? Search and reuse the owning helper, type, pattern, or seam.
  3. Does the standard library solve it? Use it.
  4. Does a native platform feature solve it? Prefer HTML, CSS, database, framework, or operating-system behavior over custom code.
  5. Does an installed dependency already solve it? Reuse it; do not add a package for a few clear lines.
  6. Can one readable local expression solve it? Keep it local.
  7. Only then: write the minimum new code that is correct.

The ladder follows comprehension, never replaces it. Read the task and repository instructions, trace the real callers and consumers, and identify the source of truth before choosing a rung.

Keep the change genuinely small

  • Fix the root cause at the shared owning seam. A one-line guard in the common path is smaller than patches in every caller.
  • Preserve every explicit requirement and reachable security, privacy, accessibility, data-integrity, error-handling, and compatibility invariant.
  • Prefer deletion over addition and boring code over clever code.
  • Avoid one-implementation interfaces, one-product factories, speculative configuration, parallel frameworks, scaffolding "for later," and new dependencies without measured need.
  • Use the fewest files that produce one coherent, readable outcome. Do not force a one-liner that hides behavior or moves complexity elsewhere.
  • When two small solutions are equivalent, prefer the one with safer edge behavior and lower ownership cost.
  • Mark a deliberate simplification with a real ceiling using a ponytail: comment that names the ceiling and upgrade trigger.

A complex request may have a simpler equivalent path. Use it and explain the substitution in one line, but never drop a requirement merely to reduce the diff.

Verify proportionately

A lazy change without proof is unfinished. Follow existing repository test conventions:

  • Non-trivial logic such as a branch, loop, parser, or money/security path gets at least one focused runnable check.
  • Add more checks only for materially distinct risks or acceptance conditions.
  • Do not introduce a test framework for one change.
  • A trivial one-liner needs no new test when existing coverage already proves it.

Set intensity

LevelBehavior
liteBuild what was asked and name the lazier equivalent in one line.
fullEnforce the ladder; prefer reuse, standard library, native behavior, and the shortest readable correct diff.
ultraChallenge speculative requirements, try deletion first, and ship only the smallest demonstrated need.

User requirements and safety invariants always outrank intensity.

Stop and ask when

  • the lazier equivalent would drop an explicit requirement or a reachable safety, privacy, accessibility, or data-integrity invariant;
  • the smallest correct change needs a consequential product, contract, or authority decision only the user can make.

Report briefly

Lead with the delivered code or outcome. Then use at most three short lines for what was skipped and the condition that would justify adding it, unless the user explicitly requested a report or walkthrough.

Pattern: [result] -> skipped: [X]; add when [Y].

The shortest correct, readable path to done is the right path.

Done when

The requested behavior works, is proven proportionately, and the report follows the pattern above. Ponytail governs what is built, not whether the task is understood or whether required evidence is reported.

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.