How do I install the Command Authoring skill?

Run npx skillmds@latest add ibrahim-3d/command-authoring 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.

Is Command Authoring safe to use?

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

Which AI agents work with Command Authoring?

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.

Is Command Authoring free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Command Authoring?

ibrahim-3d (@ibrahim-3d) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Command Authoring by ibrahim-3d — Agent Skill

Name: Command Authoring
Author: ibrahim-3d

Command Authoring

Write or refactor Claude Code slash commands (files in commands/<ns>/<cmd>.md). Use when the user asks to "create a slash command", "add /<name>", "write a command file", "make a /command", "turn this workflow into a command", "argument hints", "allowed-tools", or when packaging a repeatable workflow as a one-keystroke invocation. Covers frontmatter, $ARGUMENTS parsing, @-imports for context files, and orchestration patterns.

ibrahim-3d 480b1a4 3.2 KB Updated Sep 17, 2026

File contents

Slash Command Authoring

A slash command is a markdown file at commands/<namespace>/<name>.md. Invoked as /namespace:name [args].

Frontmatter

---
name: <namespace>:<cmd>
description: One-line what the command does. Shows in /help.
argument-hint: "[arg1] [arg2]"   # what $ARGUMENTS will contain
allowed-tools:
  - Read
  - Write
  - Bash
  - Task
  - AskUserQuestion
---

allowed-tools is an allowlist — if set, the command runs with only these. Omit to inherit the session's tools. Prefer allowlists for safety.

Body structure

<objective>
One paragraph: what the command produces and what files it touches.
</objective>

<execution_context>
@${CLAUDE_PLUGIN_ROOT}/<plugin>/references/<doc>.md
@${CLAUDE_PLUGIN_ROOT}/<plugin>/templates/<template>.md
</execution_context>

<context>
Arguments: $ARGUMENTS
Parse as: <describe>
</context>

<process>
Numbered steps. Each step is either:
- Read a file
- AskUserQuestion to fill a gap
- Task-spawn a specialist (by subagent_type)
- Write an output file
</process>

<output>
Files created/modified + a 5-line summary.
</output>

`@`-imports

@<path> in the body loads the file contents inline at command expansion time. Use ${CLAUDE_PLUGIN_ROOT} for plugin-relative paths. Use this to pull in long reference docs, templates, and few-shot examples so the command body itself stays short.

$ARGUMENTS

Raw string of everything after the command name. Parse it in the body prose — the model handles extraction. For multi-arg commands, document the expected format in argument-hint.

Orchestration commands vs. task commands

Task command: does one thing inline (e.g., /note, /status).
Orchestration command: spawns multiple specialists (e.g., /agents-architect:new, GSD's /gsd:plan-phase). Includes gates, approvals, and aggregates specialist output.

Orchestrators should always use the Task tool, not inline the specialist's work.

Design principles

Idempotent: running the command twice produces the same result. Check for existing state before creating.
Explicit gates: any non-trivial action (writing >1 file, destructive ops) gets AskUserQuestion confirmation.
Short body, deep imports: command body under 100 lines; push detail into @-imported references.
State on disk: commands that span turns write checkpoint files (.planning/STATE.md, .arch/state.json) rather than relying on context.

References

@../../agents-architect/references/anthropic-best-practices.md
@../../agents-architect/templates/command.template.md

Output

Produce commands/<ns>/<name>.md with correct frontmatter, XML-structured body, @-imports for heavy context, and a short exit summary spec.

---name: command-authoringdescription: Write or refactor Claude Code slash commands (files in commands/<ns>/<cmd>.md). Use when the user asks to "create a slash command", "add /<name>", "write a command file", "make a /command", "turn this workflow into a command", "argument hints", "allowed-tools", or when packaging a repeatable workflow as a one-keystroke invocation. Covers frontmatter, $ARGUMENTS parsing, @-imports for context files, and orchestration patterns.---# Slash Command AuthoringA slash command is a markdown file at `commands/<namespace>/<name>.md`. Invoked as `/namespace:name [args]`.## Frontmatter```yaml---name: <namespace>:<cmd>description: One-line what the command does. Shows in /help.argument-hint: "[arg1] [arg2]" # what $ARGUMENTS will containallowed-tools: - Read - Write - Bash - Task - AskUserQuestion---````allowed-tools` is an **allowlist** — if set, the command runs with only these. Omit to inherit the session's tools. Prefer allowlists for safety.## Body structure```markdown<objective>One paragraph: what the command produces and what files it touches.</objective><execution_context>@${CLAUDE_PLUGIN_ROOT}/<plugin>/references/<doc>.md@${CLAUDE_PLUGIN_ROOT}/<plugin>/templates/<template>.md</execution_context><context>Arguments: $ARGUMENTSParse as: <describe></context><process>Numbered steps. Each step is either:- Read a file- AskUserQuestion to fill a gap- Task-spawn a specialist (by subagent_type)- Write an output file</process><output>Files created/modified + a 5-line summary.</output>```## `@`-imports`@<path>` in the body loads the file contents inline at command expansion time. Use `${CLAUDE_PLUGIN_ROOT}` for plugin-relative paths. Use this to pull in long reference docs, templates, and few-shot examples so the command body itself stays short.## $ARGUMENTSRaw string of everything after the command name. Parse it in the body prose — the model handles extraction. For multi-arg commands, document the expected format in `argument-hint`.## Orchestration commands vs. task commands- **Task command**: does one thing inline (e.g., `/note`, `/status`).- **Orchestration command**: spawns multiple specialists (e.g., `/agents-architect:new`, GSD's `/gsd:plan-phase`). Includes gates, approvals, and aggregates specialist output.Orchestrators should **always** use the Task tool, not inline the specialist's work.## Design principles1. **Idempotent**: running the command twice produces the same result. Check for existing state before creating.2. **Explicit gates**: any non-trivial action (writing >1 file, destructive ops) gets `AskUserQuestion` confirmation.3. **Short body, deep imports**: command body under 100 lines; push detail into `@`-imported references.4. **State on disk**: commands that span turns write checkpoint files (`.planning/STATE.md`, `.arch/state.json`) rather than relying on context.## References- @../../agents-architect/references/anthropic-best-practices.md- @../../agents-architect/templates/command.template.md## OutputProduce `commands/<ns>/<name>.md` with correct frontmatter, XML-structured body, @-imports for heavy context, and a short exit summary spec.

Frequently asked questions