Feedback on your fetching-library-docs skill

[RichardHightower]

I checked out the fetching-library-docs skill and noticed you're tackling the documentation retrieval problem with a pretty thoughtful approach to handling different library formats—curious whether you've run into cases where the extraction logic needs to adapt for libraries with non-standard doc structures, or if that's something you're planning to handle as it comes up?

Links:

• GitHub
• Your agentic skill in the SkillzWave marketplace

The TL;DR

You're at 85/100, solid B-grade territory. This is based on Anthropic's skill grading standards. Your strongest area is Writing Style (9/10)—the technical voice is clean and consistent. Weakest area is PDA/Progressive Disclosure (23/30)—you've got good layering but both your main file and reference doc exceed 200+ lines without navigation aids, which burns tokens unnecessarily.

What's Working Well

• Strong trigger phrases in the metadata description ('how do I use X library', 'code examples')—this will help people find it when they need it
• Clear scope boundaries—the "NOT for" exclusion (repo internals) tells users exactly when NOT to use this, which is smart
• Solid feedback loops—you've got verification steps, a --verbose flag for token stats, and a troubleshooting table with stop-and-report patterns baked in
• Real token efficiency claims—77% reduction is measurable and actually useful, not marketing fluff

The Big One: Navigation in Long Files

Your SKILL.md hits 222 lines and the reference doc hits 226 lines, but neither has a Table of Contents. This is burning tokens for anyone trying to find what they need.

Fix: Add a ToC at the top of both files:

## Table of Contents
- [Quick Start](#quick-start)
- [Standard Workflow](#standard-workflow)
- [Parameters](#parameters)
- [Workflow Patterns](#workflow-patterns)
- [Troubleshooting](#troubleshooting)

This nets you +2 points on PDA and makes the skill actually scannable.

Other Things Worth Fixing

1. Token savings repetition—You mention "77% reduction" three separate times (lines 14, 30, 161). Pick one anchor point (Quick Start result) and reference it elsewhere. +1 point

2. Missing output examples—Your workflow patterns show bash commands but no sample output showing what successful responses actually look like. Add one real output block so people know what to expect. +1 point

3. Consolidate patterns—You've got 4 workflow patterns that are structurally similar. Trim to 2-3 core scenarios (Basic Query, Multi-Step Learning, Pagination) with variation notes. Cuts ~15 lines and tightens PDA. +1 point

Quick Wins

• Add ToC to both files (+2 PDA)
• Reduce "77% savings" mentions to one location (+1 Writing Style)
• Include sample output blocks in workflow patterns (+1 Utility)
• Consolidate 4 patterns into 2-3 core scenarios (+1 PDA)

These changes would push you from 85 to 89-90/100 (A-minus territory).

---

Checkout your skill here: [SkillzWave.ai](https://skillzwave.ai) | [SpillWave](https://spillwave.com) 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.