personalization-logic

I found your personalization-logic skill while reviewing the latest batch—the approach to handling user preference trees is clever, but I'm seeing some gaps in how the spec translates into actual implementation that's dragging your score down to 68.

Links:

• GitHub
• Your agentic skill in the SkillzWave marketplace

The TL;DR

You're at 68/100, which puts you in D territory—needs work but fixable. This is based on Anthropic's skill best practices rubric. Your strongest pillar is Spec Compliance (12/15), but you're getting dinged hard on Progressive Disclosure Architecture (18/30) and Utility (12/20). The gap between a solid framework and a usable skill is mostly missing concrete examples and templates.

What's Working Well

• Your framework is solid — the five-step flow (Segmentation Inputs → Content Blocks → Token Logic → Fallbacks → Testing) maps real personalization needs and shows you've thought through the domain.
• Spec compliance is tight — valid YAML frontmatter, correct naming convention (hyphen-case), and you're using consistent terminology throughout ("segment," "token," "module"). No conflicts there.
• Objectivity is clean — the tone stays instructional without marketing fluff, which is refreshing and appropriate for something this technical.

The Big One: Missing Templates and Examples

This is eating ~8 points and it's the easiest win. You mention three templates in your "Templates" section—personalization matrix, token dictionary, QA checklist—but they don't actually exist as reference files. You're telling people what to build without showing how.

Fix: Create three new files in a references/ directory:

• references/personalization-matrix-template.md — actual table showing segment vs. module vs. asset mappings
• references/token-dictionary-template.md — example tokens with field names, data sources, fallbacks
• references/qa-checklist.md — checklist template for testing

This pushes you into PDA's "Layered Structure" and gives you proper depth. You'll gain those 8 points just by providing what you're already describing.

Other Things Worth Fixing

1. Add a workflow procedure (Ease of Use, +4 points) — Your Framework section lists components but no numbered steps. Reframe as: "1. Extract segmentation inputs from CRM. 2. Map segments to content blocks. 3. Define tokens with data sources. 4. Configure fallbacks. 5. Validate with seed records."

2. Zero concrete examples (Utility, +3 points) — Include actual personalization rules: IF company.industry="SaaS" THEN hero_block="cloud-first-message" or show token syntax like {{company.industry | fallback: "Enterprise"}}.

3. Vague dependency docs (Ease of Use, +2 points) — You mention "upstream data hygiene" but don't specify what that means. Say: "Required fields: company.industry (standardized), user.role (normalized), account.lifecycle_stage (validated). Require >95% data completeness."

4. Missing validation steps (Utility, +3 points) — Your "Testing Plan" talks about A/B tests but no quality gates. Add: "Validation: Run seed list test → Check fallback rendering → Verify token resolution → Confirm segment coverage."

Quick Wins

• Add reference files for the three templates you're already describing (+8 points)
• Convert framework into a numbered workflow (+4 points)
• Drop in one or two concrete examples of personalization rules (+3 points)
• Specify data dependencies clearly instead of hand-waving (+2 points)

That's +17 points by addressing the gap between your solid conceptual framework and the actionable implementation details people actually need. You go from D to solid C territory.

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.