commander

I was curious how you'd approach the CLI command parsing problem—commander really leans into chainable APIs, and at 80 points there's something interesting about the structure that caught my attention but also some gaps worth exploring.

Links:

• GitHub
• Your agentic skill in the SkillzWave marketplace

The TL;DR

You're at 80/100, solid B territory. This is based on Anthropic's Claude Skills best practices. Your writing style is your strongest asset (9/10)—clean, imperative, zero fluff. On the flip side, Progressive Disclosure Architecture (24/30) and Ease of Use (22/25) are where the points are leaving the table. You've got the fundamentals down, but the structure and guidance could be tighter.

What's Working Well

• Token economy is tight: Every line does work. No marketing language, no padding. That's how you keep developers in your skill efficiently.
• Trigger phrases nail the discovery problem: "Commander.js, CLI, commands, options"—someone searching for command-line patterns will find this immediately.
• Code examples are strong: Your table format for patterns makes them quick to scan. That's the kind of thing that keeps devs from leaving halfway through.
• Consistency with terminology: You stick with "command," "option," "action" throughout. That reduces cognitive load.

The Big One: Missing Progressive Disclosure

The entire skill lives in a single 88-line SKILL.md file. That's lean, but you're leaving structure on the table. Here's what's happening: advanced patterns (error handling, git-style subcommands, plugin systems) could live in a references/advanced.md without bloating the main file. This hits +2-3 points on PDA and makes the skill more scalable.

The structure would look like:

.claude/skills/commander/
├── SKILL.md (40-50 lines, core patterns on...