mdr-745-specialist

Found your mdr-745-specialist skill while reviewing the grading pipeline—the approach to specialist routing caught my eye, especially given the 68/100 score suggests some solid fundamentals but room to tighten the architecture. What's driving the focus on MDR-specific handling here?

Links:

• GitHub
• Your agentic skill in the SkillzWave marketplace

The TL;DR

You're at 68/100, solid D-grade territory. The skill has strong fundamentals—your Spec Compliance (12/15) and metadata are legit—but you're losing major points on Progressive Disclosure Architecture (17/30) and Utility (13/20). The core issue: you're front-loading too much detail into SKILL.md when it should be spread across reference files, and your promised implementations (scripts, templates, detailed guides) don't actually exist yet.

What's Working Well

• Spec Compliance is locked in — Your frontmatter is valid YAML, naming follows conventions perfectly (hyphen-case), and you've got solid trigger phrases like "MDR compliance," "classification," and "technical documentation"
• Clear workflow structure — The numbered decision trees (Classification Determination, Gap Analysis Framework, Documentation Preparation) show you understand the problem space and have thought through the process
• Good navigational signals — Even at 196 lines, your headers make it scannable. The "when to use" section is practical

The Big One: Reference Files Are Placeholders

This is eating ~5 points right now. You reference detailed guides (mdr-classification-guide.md, clinical-evidence-requirements.md, technical-documentation-templates.md) throughout SKILL.md, but references/api_reference.md is just a stub:

# Reference Documentation for Mdr 745 Specialist
This is a placeholder for detailed reference documentation.

Here's the fix: Move the heavy detail out of SKILL.md and actually build those reference files. Take sections like "Technical Documentation Structure" (currently 20+ lines in SKILL.md) and move them to references/technical-documentation-templates.md. Then in SKILL.md, just say: "See references/technical-documentation-templates.md for the complete structure." This shrinks SKILL.md, improves token economy, and makes you searchable without sacrificing comprehensiveness.

Impact: +5 points if executed properly.

Other Things Worth Fixing

1. Scripts and assets don't exist — You mention mdr-gap-analysis.py and clinical-evidence-tracker.py but only have example.py. Either implement them or remove the specific file references. Vague is better than broken promises.

2. Zero concrete examples — No sample classification rationales, no sample gap analysis output, no actual documentation section templates. Add 2-3 worked examples showing input → classification logic → output. This alone bumps Utility up ~3 points.

3. Marketing language in the opener — "Expert-level...comprehensive knowledge" feels sales-y. Replace with objective capabilities: "EU MDR 2017/745 compliance specialist covering device classification, technical documentation, clinical evidence requirements, and post-market surveillance."

4. Missing validation steps — Your Gap Analysis Framework describes what to assess but doesn't define completion criteria or verification checkpoints. Add explicit "How do you know the analysis is complete?" checks.

Quick Wins

• Most impactful: Build out reference files with actual content from SKILL.md (30 min work, +5 points)
• Next: Add 2-3 concrete examples showing real device classifications and their rationale (+3 points)
• Then: Implement or remove promised scripts; don't reference tools that don't exist (+4 points)
• Polish: Trim marketing language, add validation steps (+2 points combined)

You're well-positioned at 68—another 15-20 points is totally achievable by moving from "promises detail" to "has detail available where it matters."

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.