community-insight-taxonomy

I checked out your community-insight-taxonomy skill and there's some solid foundational thinking here—the taxonomical approach to categorizing insights is genuinely useful. That said, the 58/100 score suggests there are some gaps between the concept and execution that might be worth exploring, particularly around how discoverable and actionable the taxonomy actually is for developers using it.

Links:

• GitHub
• Your agentic skill in the SkillzWave marketplace

TL;DR

You're at 58/100, solidly in F territory. The grading is based on Anthropic's five-pillar rubric for agentic skills. Writing style is your strongest pillar (9/10—clear, objective prose) but Progressive Disclosure Architecture is dragging you down (13/30)—you're cramming everything into 32 lines with no supporting reference files, and Ease of Use (14/25) suffers because there's no actionable workflow, just a list of dimensions.

What's Working Well

• Writing voice is tight. You're using imperative/infinitive tone correctly throughout, zero marketing fluff. Graders noticed that right away.
• Terminology is consistent. "Persona," "journey," "driver," "sentiment"—these terms track cleanly through the entire skill with no conflicting definitions.
• Structure is clear. Your headers are well-organized and the content flows logically from framework → tips → integration points.

The Big One: Missing Reference Files

This is your biggest opportunity. Your Templates section promises three things ("Tag dictionary with definitions," "Airtable/Sheet schema," "Quick-reference cheatsheet") but none of these files actually exist. This tanks your PDA score (Progressive Disclosure) because you're not layering your content—you're just talking about layers.

The fix: Create three reference files in a references/ subdirectory:

• references/tag-dictionary.md – Define each persona, journey stage, driver, and sentiment value with examples
• references/schema-template.md – Provide an actual Airtable/Sheet template structure
• references/cheatsheet.md – A one-page moderator's quick-reference

This alone could bump you +10 points toward 70+.

Other Things Worth Fixing

1. Add trigger phrases to the description. Right now it reads like "Tagging schema for classifying community signals..." but graders are looking for action verbs. Change it to include "Use when asked to 'classify community signals', 'tag feedback by persona', or 'structure insights'." (+2-4 points)

2. Add a numbered workflow. Your Framework lists dimensions but doesn't tell developers how to actually tag something step-by-step. Add: 1. Read signal 2. Identify persona 3. Determine journey stage 4. Assign driver(s) 5. Rate sentiment 6. Link to business metric. (+5 points)

3. Include validation logic. You mention "monthly taxonomy reviews" in Tips but provide no actual validation workflow. Add a section on checking inter-rater reliability, flagging low-confidence tags, or resolving classification ambiguities. (+4 points)

Quick Wins

• Create the three reference files (biggest impact: ~+10 points)
• Rewrite description with specific trigger phrases (+2-4 points)
• Add numbered workflow steps to Framework (+5 points)
• Include a validation/feedback section (+4 points)

Targeting these four items could realistically get you to 75-80/100. The foundational concept is there—you just need to flesh out the supporting structure and make it more actionable for developers.

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.