How do I install the Help skill?

Run npx skillmds add oyi77/help in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

What does the Help skill do?

Use when quick reference for ponytail modes, skills, and commands. One-shot display. It is listed under Coding & Dev Tools on SkillMD.

This skill has not completed SkillMD's automated safety review yet. Capability flags: docs only. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Help?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Yes. Installing skills from SkillMD is free. This skill is licensed under Apache-2.

oyi77 (@oyi77) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Help by oyi77 — Agent Skill

Name: Help
Rating: 4.2 (5 reviews)
Author: oyi77

File contents

Help — Quick Reference Card for Ponytail Modes

Quick Reference

One-shot display of ponytail modes, key commands, and skill discovery. Never changes mode, writes files, or persists anything. Pure reference.

Overview

Help mode prints a concise reference card covering all four ponytail modes (audit, debt, review, help) and their commands, options, and output formats. It is the fastest way to learn the tools without reading the full parent skill. Designed for the moment you know what you want to do but cannot remember the exact flag.

Scope: session only. Output: reference card. Best outcome: developer finds the answer in 5 seconds.

Quick Start

Show help: ponytail help — prints the full reference card
Filter by mode: ponytail help --mode audit — shows only audit commands
Show examples: ponytail help --examples — includes usage examples for each mode

Key Command

ponytail help          # full reference card
ponytail help audit    # specific mode details
ponytail help --json   # machine-readable output for tooling

Reference Card Output

~ ponytail commands ~
  audit    — Ranked list of what to delete/simplify repo-wide
  debt     — Collect ponytail: comments into a tracked ledger
  help     — This reference card
  review   — One-line-per-finding diff over-engineering check

Options:
  --path <dir>      Scope to directory
  --diff <text>     Diff text for review
  --output <file>   Write results to file
  --stdin           Read input from stdin
  --scan            Scan mode for debt
  --rank-by-impact  Sort audit findings by LOC savings
  --require-no-findings  Fail review if any issues found

Help Checklist

Reference card displays all four modes with correct descriptions
All command options are listed and accurate
Mode filtering works (show only one mode's commands)
Examples are up-to-date with current tool behavior
Output is parseable (plain text or --json)

When to Use

Use when quick reference for ponytail modes, skills, and commands. One-shot display.

Workflow

Execute these steps sequentially:

Show help: ponytail help — prints the full reference card
Filter by mode: ponytail help --mode audit — shows only audit commands
Show examples: ponytail help --examples — includes usage examples for each mode

Anti-Rationalization Table

Rationalization	Reality
"I'll just read the source"	Help is faster than grepping source. It is the single entry point — use it first
"I already know the commands"	Flags change, options get added. A 5-second help check prevents stale assumptions
"I can guess the flag names"	`--rank-by-impact` vs `--sort-by-size` — guessing wastes more time than reading help once

Process

Invoke ponytail help for full reference
Filter by mode if needed: --mode audit|debt|review
Show examples with --examples for usage patterns
Apply relevant mode to current task

Verification

Help output displays without errors
All modes listed with correct descriptions
Examples match current CLI behavior
Output parseable with --json flag

---name: helpdescription: Use when quick reference for ponytail modes, skills, and commands. One-shot display.license: Apache-2.0---# Help — Quick Reference Card for Ponytail Modes## Quick ReferenceOne-shot display of ponytail modes, key commands, and skill discovery. Never changes mode, writes files, or persists anything. Pure reference.## OverviewHelp mode prints a concise reference card covering all four ponytail modes (audit, debt, review, help) and their commands, options, and output formats. It is the fastest way to learn the tools without reading the full parent skill. Designed for the moment you know what you want to do but cannot remember the exact flag.Scope: session only. Output: reference card. Best outcome: developer finds the answer in 5 seconds.## Quick Start1. **Show help**: `ponytail help` — prints the full reference card2. **Filter by mode**: `ponytail help --mode audit` — shows only audit commands3. **Show examples**: `ponytail help --examples` — includes usage examples for each mode## Key Command```bashponytail help # full reference cardponytail help audit # specific mode detailsponytail help --json # machine-readable output for tooling```## Reference Card Output```~ ponytail commands ~ audit — Ranked list of what to delete/simplify repo-wide debt — Collect ponytail: comments into a tracked ledger help — This reference card review — One-line-per-finding diff over-engineering checkOptions: --path <dir> Scope to directory --diff <text> Diff text for review --output <file> Write results to file --stdin Read input from stdin --scan Scan mode for debt --rank-by-impact Sort audit findings by LOC savings --require-no-findings Fail review if any issues found```## Help Checklist- [ ] Reference card displays all four modes with correct descriptions- [ ] All command options are listed and accurate- [ ] Mode filtering works (show only one mode's commands)- [ ] Examples are up-to-date with current tool behavior- [ ] Output is parseable (plain text or --json)## When to UseUse when quick reference for ponytail modes, skills, and commands. One-shot display.## WorkflowExecute these steps sequentially:1. **Show help**: `ponytail help` — prints the full reference card2. **Filter by mode**: `ponytail help --mode audit` — shows only audit commands3. **Show examples**: `ponytail help --examples` — includes usage examples for each mode## Anti-Rationalization Table| Rationalization | Reality ||---|---|| "I'll just read the source" | Help is faster than grepping source. It is the single entry point — use it first || "I already know the commands" | Flags change, options get added. A 5-second help check prevents stale assumptions || "I can guess the flag names" | `--rank-by-impact` vs `--sort-by-size` — guessing wastes more time than reading help once |## Process1. **Invoke** `ponytail help` for full reference2. **Filter** by mode if needed: `--mode audit|debt|review`3. **Show examples** with `--examples` for usage patterns4. **Apply** relevant mode to current task## Verification- Help output displays without errors- All modes listed with correct descriptions- Examples match current CLI behavior- Output parseable with `--json` flag

Frequently asked questions