Antigravity Skills - Build and Discover AI Agent SkillsAntigravity Skills

Ponytail Skill · Coding discipline

Ponytail Skill: Stop Overbuilding. Ship the Smallest Correct Change.

Read the real flow. Climb only as far as the requirement needs.

Ponytail keeps explicit requirements as the floor, reuses the existing code path, and stops at the first ladder rung that fully satisfies the task. The seven-rung discipline remains intact.

Source diff verifiedImmutable package verified
Coding disciplineMinimal changesYAGNI
View canonical source
0
0

About This Skill

Ponytail is an AI coding-agent Skill that favors the smallest correct implementation after tracing the behavior and preserving explicit safety, data-loss, accessibility, and validation requirements.

Source and Release

Provenance

Upstream author
Dietrich Gebert
Revision
16f29800fd2681bdf24f3eb4ccffe38be3baec6b
Upstream version
4.8.4 + main@16f2980
License
MIT

Antigravity release

Version
0.300.1-885a6c0051e8
Package hash
sha256:bb433a30bbfa4cc1ad89a75b390b0d315e830aa581bed08a64b8597a3e8cd885
Package status
approved
Files
1
ZIP size
2.1 KB

Changelog

  • Deterministic ZIP built from the optimized package with a versioned SHA-256 artifact.

What We Improved

Read, then climb

The candidate makes the order explicit: trace every affected path first, then stop at the earliest ladder rung that satisfies the requirement.

Requirements remain the floor

Minimality cannot erase validation, data-loss protection, security, accessibility, or explicit behavior requested by the user.

Guardrails sit beside the coding decision

Present need, validation, data-loss, security, accessibility, edge-case, and clarification rules are co-located with the implementation choices they constrain.

A six-item completion gate

The final handoff checks requested behavior, affected callers, validation, necessary-now status, and the selected mode before stopping.

Quick Start

  1. 1State the requested behavior, affected surface, and explicit constraints.
  2. 2Trace every caller and code path that can be affected before choosing an implementation.
  3. 3Climb the seven-rung ladder from YAGNI and reuse through native or installed capabilities only as needed.
  4. 4Run the focused check, verify affected callers and guardrails, then stop at the first correct rung.

Core Capabilities

Seven-rung ladder

YAGNI, reuse, standard library, native platform, installed dependency, one expression, and minimum implementation remain ordered.

Real-flow tracing

The Skill asks the agent to trace callers and affected paths before it chooses the smallest change.

Guardrails stay visible

Safety, validation, data-loss, security, accessibility, and edge-case handling remain explicit floors.

Checkable stop condition

The completion gate makes “done” observable instead of equating smallness with correctness.

Use Cases

Bug fixes

Trace the failing boundary and callers, then fix the shared root cause once.

Small feature work

Reuse the project helper and native capability before introducing a new layer or dependency.

Refactors and reviews

Keep the requested behavior and stop when the smallest correct change is verified.

Source diff and package proof verified

Evidence: source diff

Verified source-diff evidence

This release proves canonical source lineage, package inventory, the optimized true diff, and an immutable artifact. It does not claim runtime outcomes or visual comparisons.

  • .scratch/kan-300/production-release/source-evidence/candidates/ponytail/original-inventory.json#/entries?sha256=8271b5e04f0fa3bbb9e30410b3b3bb2f0f4d3d3092aeb24817c2084ed5a1de84
  • .scratch/kan-300/production-release/source-evidence/candidates/ponytail/optimized-inventory.json#/entries?sha256=bddc963733ce8092c43903de19d82725566406eb29b39aa1fe26132bf9de81ac
  • .scratch/kan-300/production-release/source-evidence/candidates/ponytail/change-set.json#/evidenceUnits/0?sha256=ec0f62ef2211f7d339e690e94fc6281d8c653476c87496d1e5fb2b90412dd345
  • .scratch/kan-300/production-release/source-evidence/candidates/ponytail/change-set.json#/evidenceUnits/1?sha256=ec0f62ef2211f7d339e690e94fc6281d8c653476c87496d1e5fb2b90412dd345
  • .scratch/kan-300/production-release/source-evidence/candidates/ponytail/change-set.json#/evidenceUnits/2?sha256=ec0f62ef2211f7d339e690e94fc6281d8c653476c87496d1e5fb2b90412dd345
  • .scratch/kan-300/production-release/source-evidence/candidates/ponytail/change-set.json#/evidenceUnits/3?sha256=ec0f62ef2211f7d339e690e94fc6281d8c653476c87496d1e5fb2b90412dd345
Verification receipt
.scratch/kan-300/production-release/source-evidence/candidates/ponytail/change-set.json?sha256=ec0f62ef2211f7d339e690e94fc6281d8c653476c87496d1e5fb2b90412dd345
Source-diff receipt
.scratch/kan-300/production-release/source-evidence/candidates/ponytail/change-set.json?sha256=ec0f62ef2211f7d339e690e94fc6281d8c653476c87496d1e5fb2b90412dd345

SKILL.md

# Ponytail

This edition derives from DietrichGebert/ponytail at 16f29800fd2681bdf24f3eb4ccffe38be3baec6b; its source snapshot is labeled 4.8.4 + main@16f2980.

## Read, then climb

Trace the real flow and explicit requirements first. Stop at the first ladder rung that fully satisfies them: YAGNI, reuse, standard library, native platform, installed dependency, one expression, or minimum implementation.

The one-file Skill keeps its MIT notice boundary and the six-item completion gate. This release edition has no matched coding-task transcript.

Related Skills

Superpowers SkillsA disciplined software development workflow.Caveman SkillTerse replies that keep technical signal.

Frequently Asked Questions

FAQ

What is Ponytail Skill?
Ponytail is a coding-agent discipline for understanding the task and real code path before choosing the smallest correct implementation.
Does Ponytail mean fewer characters at any cost?
No. Explicit requirements and trust-boundary safeguards remain the floor; minimality applies only after those requirements are understood.
What are the seven rungs?
The candidate preserves YAGNI, reuse, standard library, native platform, installed dependency, one expression, and minimum implementation in that order.
Is the smaller-change behavior proven?
Not yet. A matched coding-task workflow and reviewer receipt are required; no token, cost, speed, or safety uplift is claimed here.
Can I download it now?
Yes. Download the approved immutable Ponytail Skill: Minimalist Coding Discipline for AI Agents package from this page. Its source revision, file count, byte size, SHA-256, and release receipt are shown here; this evidence does not claim runtime outcomes.

Discussion

Discussion

0 comments
U

Quick Info

Author
Dietrich Gebert
Published
Aug 1, 2026
Category
Software development
Version
0.300.1-885a6c0051e8

Tags

Coding disciplineMinimal changesYAGNI

On This Page

Trigger Phrases

Use these phrases to activate this skill in your AI coding assistant:

use Ponytailminimal solutionstop over-engineeringsmallest correct changelazy senior developer mode