What Slash Commands Does Ponytail Offer and What Are Their Purposes?

Ponytail exposes six slash commands—/ponytail, /ponytail-review, /ponytail-audit, /ponytail-debt, /ponytail-gain, and /ponytail-help—that inject specialized skill prompts into a Hermes-compatible gateway to reduce over-engineering and track technical debt.

Ponytail is a Hermes-compatible plugin that acts as a "lazy senior dev" assistant for analyzing codebases. If you are wondering what slash commands Ponytail offers to streamline your workflow, this guide covers the complete command set defined in the DietrichGebert/ponytail repository and their specific purposes as implemented in the source code.

Available Slash Commands in Ponytail

The plugin registers all commands in the SKILL_COMMANDS dictionary located in ponytail/main/__init__.py (lines 14–20). Each command maps to a specific skill prompt that the LLM agent processes when invoked through a compatible gateway.

/ponytail — Toggle Runtime Modes

The generic /ponytail command controls the plugin's operational intensity. According to the source code, it accepts arguments lite, full, ultra, or off to set the active mode. The _handle_mode_command function (lines 66–78) processes these arguments to adjust how aggressively Ponytail analyzes your code.

/ponytail-review — Review Code for Over-Engineering

Use /ponytail-review to analyze "the current diff or provided target for over-engineering." This command injects the ponytail:review skill prompt, which evaluates whether your changes are unnecessarily complex compared to simpler alternatives.

/ponytail-audit — Audit Repository for Bloat

The /ponytail-audit command triggers a full repository scan. Its purpose is to "audit the repo for over-engineering and deletion opportunities," helping identify dead code, redundant abstractions, or architectural overstretch that can be safely removed.

/ponytail-debt — Track Shortcut Upgrade Paths

When you need to document technical debt deliberately introduced for speed, /ponytail-debt lists "every deliberate ponytail: shortcut and its upgrade path." This maintains a concrete roadmap for future refactoring when priorities shift from velocity to maintainability.

/ponytail-gain — Measure Impact Scoreboard

The /ponytail-gain command displays "the measured-impact scoreboard (less code, less cost, more speed)." It quantifies the benefits of removing over-engineered solutions, providing concrete metrics on code reduction and performance improvements.

/ponytail-help — Command Reference

Finally, /ponytail-help provides inline documentation by showing "the Ponytail command reference," making it easy to recall available options without leaving your chat interface or IDE.

How Ponytail Registers Slash Commands

The command registration happens in the register function within ponytail/main/__init__.py (lines 95–112). This function binds each entry from the SKILL_COMMANDS dictionary to the Hermes gateway context. When a user sends a slash command, the plugin intercepts the message and rewrites it into the corresponding skill prompt defined in the respective SCILL.md files located under ponytail/main/skills/.

Using Slash Commands in Practice

You invoke these commands through any Hermes-compatible gateway. The following examples demonstrate how each command triggers its respective skill prompt:

# Switch Ponytail to lite mode

gateway.receive_message("/ponytail lite")

# Response: "Ponytail mode set to lite."

# Request code review of current diff

gateway.receive_message("/ponytail-review")

# Injects the ponytail:review skill prompt

# Audit repository for over-engineering

gateway.receive_message("/ponytail-audit")

# Injects the ponytail:audit skill prompt

# List technical debt and upgrade paths

gateway.receive_message("/ponytail-debt")

# Injects the ponytail:debt skill prompt

# View impact metrics

gateway.receive_message("/ponytail-gain")

# Injects the ponytail:gain skill prompt

# Display help reference

gateway.receive_message("/ponytail-help")

# Injects the ponytail:help skill prompt

Summary

• Ponytail provides six slash commands accessed via a Hermes-compatible gateway.
• The /ponytail command adjusts runtime modes (lite, full, ultra, off) via _handle_mode_command (lines 66–78).
• Analysis commands (/ponytail-review, /ponytail-audit) identify over-engineering in diffs and repositories.
• Tracking commands (/ponytail-debt, /ponytail-gain) document shortcuts and measure code reduction impact.
• All commands are defined in SKILL_COMMANDS (lines 14–20) and registered in the register function (lines 95–112) of ponytail/main/__init__.py.
• Each command triggers a specific skill prompt documented in corresponding SCILL.md files under ponytail/main/skills/.

Frequently Asked Questions

What arguments can I pass to the /ponytail command?

The /ponytail command accepts four mode arguments: lite, full, ultra, or off. These values control the intensity of Ponytail's analysis, allowing you to toggle between minimal overhead and comprehensive auditing as implemented in the _handle_mode_command function.

How does Ponytail process slash commands internally?

When you send a slash command, the Hermes gateway routes it to the register function in ponytail/main/__init__.py (lines 95–112). The plugin matches your input against the SKILL_COMMANDS dictionary and injects the corresponding skill prompt from the ponytail/main/skills/ directory into the LLM context.

Can I use Ponytail slash commands without a Hermes gateway?

No. Ponytail is designed specifically as a Hermes-compatible plugin. The slash command interface requires a Hermes gateway (such as a chat interface) to function, as the commands rely on the gateway's receive_message method and context registration system.

Where are the skill definitions stored for each command?

Each slash command maps to a skill documented in a SCILL.md file within ponytail/main/skills/. For example, /ponytail-review uses ponytail/main/skills/ponytail-review/SCILL.md, while /ponytail-audit references ponytail/main/skills/ponytail-audit/SCILL.md.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}