Feedback on your image_gen skill

[RichardHightower]

I took a look at your image_gen skill and wanted to share some thoughts.

Links:

• GitHub
• SkillzWave Marketplace

TL;DR

You're at 86/100, solid B-grade territory. This is based on Anthropic's skill best practices rubric. Your strongest area is Ease of Use (22/25) — the workflow is crystal clear and your metadata triggers are spot-on. Weakest area is Progressive Disclosure (22/30) — you've got some redundancy between your main file and references that's eating up points.

What's Working Well

• Metadata is locked in. Your trigger phrases ('generate images', 'create cover image', 'make article illustrations') are specific and discoverable. This is why the Ease of Use score is solid.
• The 5-step workflow hits. Numbered steps with explicit outputs and checklists make it genuinely usable. Someone can follow this without guessing.
• Good examples throughout. You've got concrete cover vs. in-article image prompts, comparison tables, and input/output patterns. That's the kind of specificity that actually helps people.
• Consistent terminology. 'Cover image' vs. 'in-article image' distinction is maintained everywhere, which prevents confusion.

The Big One: Reference File is Too Thin

Your references/prompting_guide.md is only 80 lines but you're promising "comprehensive prompt engineering guidance" in SKILL.md. Right now it duplicates what's already in your main file (specificity, style, atmosphere, photography terms). This redundancy tanks your Progressive Disclosure score.

The fix: Either expand the reference with concrete examples of failed vs. refined prompts, troubleshooting strategies for bad generations, or negative prompt patterns — or consolidate it into SKILL.md and drop the separate reference. Right now you're splitting attention without gaining token efficiency. This alone could net you +3 points.

Other Things Worth Fixing

1. Missing error handling workflow. Step 5 only verifies success. Add a troubleshooting section: what happens when generations flop, how to refine prompts iteratively, retry strategies. Users will need this. (+2 points)

2. Imagen CLI setup is assumed. You mention it but don't explain how to install or verify it's working. Add a quick prerequisites check command in Overview. (+2 points)

3. Occasional second-person creep. A few instances of "your article" instead of imperative form ("Analyze the article before generating..."). Small consistency fix. (+1 point)

Quick Wins

• Expand or consolidate the reference file (biggest ROI)
• Add troubleshooting + retry patterns to Step 5
• Clarify imagen CLI prerequisites
• Tighten voice to consistent imperative form

Hit these and you're solidly in A territory (92-95 range). The core skill is genuinely useful — it just needs the supporting structure tightened up.

---

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.