article-extractor

I took a look at your article-extractor skill and wanted to share some thoughts.

Links:

• GitHub
• Your agentic skill in the SkillzWave marketplace

The TL;DR

You're at 70/100, solidly in C territory. The skill's built on solid fundamentals—your Utility scores a perfect 20/20, which means this actually solves a real problem with good degrees of freedom. But Progressive Disclosure Architecture is dragging you down at 13/30. Based on Anthropic's best practices for agentic skills, this is really about structure and navigation efficiency.

What's Working Well

• Utility is chef's kiss — You've nailed the core value prop. The skill addresses a clear need (clean article extraction), handles errors properly, and includes validation steps. That's the hard part.
• Ease of Use is nearly perfect (24/25) — Your metadata, trigger phrases, and workflow clarity are all there. Developers will find and understand this easily.
• Examples and templates are solid — You've included actual code patterns and complete workflows, not just theory.

The Big One: Layer Your Content

This is what's holding you back most. Right now you've got a 370-line SKILL.md file with everything crammed in—installation, methods, error handling, examples, all inline. That works at small scale, but it burns through tokens fast and makes navigation painful.

Here's the fix: Break it into a proper reference structure:

• Keep SKILL.md as a ~80-100 line overview: metadata, quick description, 1-2 usage examples, and links to detailed docs
• Move detailed installation to references/installation.md
• Move each method (Puppeteer, Cheerio, Playwright, etc.) to its own file: references/method-puppeteer.md, etc.
• Move error handling and edge cases to references/troubleshooting.md

This should bump you +7 points on PDA alone because you're now token-efficient and actually navigable.

Other Things Worth Fixing

1. Add a table of contents — Your sections are well-structured, but a 370-line file needs a ToC with anchor links at the top. Quick win, costs almost nothing, helps navigation (+2 points).

2. Expand your description trigger phrases — You mention "extract articles" and "web content", but add more: "scrape text from URLs", "parse HTML", "clean article content". Better discoverability.

3. Fix a few metadata quirks — The grading report flagged some technical issues (Windows paths formatting, XML tags in metadata, allowed_tools as string instead of array). These are small but knock a few points off Spec Compliance.

Quick Wins (in order of bang for buck)

• Refactor into layered structure with reference files (+7 points)
• Add table of contents (+2 points)
• Expand description trigger phrases (+1-2 points)
• Clean up metadata formatting issues (+2-3 points)

You're close to B territory (80+) with about 12-15 points of work. The skill itself is genuinely useful; it just needs better packaging.

Checkout your skill here: SkillzWave.ai | SpillWave We have an agentic skill installer that install skills in 14+ coding agent platforms. Check out this guide on how to improve your agentic skills.