Prompts, Skills, and Commands

HeroPrompt ships three kinds of content. They look similar but solve different problems. Pick the right tool first, write less re-prompting later.

Prompts

A prompt is a single, parameterised template you paste into any LLM chat.

• One shot, one purpose.
• Variables like {{topic}} get filled in before you send.
• Stays in your clipboard or the web app — does not install anywhere.

Use a prompt when the task is "write me an X" and you finish in one round-trip.

Skills

A skill is a multi-step workflow that an agent follows when it detects you need it.

• Triggered automatically by the agent based on the skill description.
• Can have branching steps, decision trees, and supporting resources.
• Installs as ~/.claude/skills/<name>/SKILL.md (with optional supporting files).

Use a skill when the work has phases — explore, plan, implement, verify — and you want the agent to follow that order without you reminding it.

Slash commands

A slash command is a callable shortcut you type in your agent chat (e.g. /commit).

• You invoke it explicitly with /<name> arguments.
• It's a markdown body ending in $ARGUMENTS that the agent runs as a prompt.
• Installs as ~/.claude/commands/<name>.md.

Use a slash command when the task is small, repetitive, and you want it ready under a keystroke.

When to use what