Grading Criteria

SkillThis grades every skill on six criteria, each with a specific weight. The total score (0-100) maps to a letter grade from A+ to F.

The Six Criteria

1. Format Compliance (15%)

Does the skill have valid structure and metadata?

Check	Requirement
YAML frontmatter	Present with name and description fields
Name format	Kebab-case, preferably gerund form
Description person	Third person only (no “I”, “You”, “We”)
Description content	Includes what it does AND trigger phrases

Critical deduction: First or second person in description = -10 points.

2. Conciseness (15%)

Does the skill respect the AI’s intelligence?

The AI already knows what common concepts are. A skill about code review doesn’t need to explain what a pull request is. A skill about data analysis doesn’t need to define what SQL is.

Good (concise):

Use pdfplumber for text extraction:
\`\`\`python
import pdfplumber
with pdfplumber.open("file.pdf") as pdf:
    text = pdf.pages[0].extract_text()
\`\`\`

Bad (verbose):

PDF (Portable Document Format) files are a common file format
that contains text, images, and other content. To extract text
from a PDF, you'll need to use a library...

Critical deduction: Over-explaining basics = -10 points.

3. Quick Start Quality (15%)

Does the skill provide immediate value upfront?

The Quick Start should be the first section after the frontmatter. It should contain a working example or a concise step-by-step that gets results immediately.

Critical deduction: No Quick Start or immediate actionable content = -15 points.

4. Workflow Quality (15%)

Does the skill have a clear step-by-step process?

Good workflows have:

• Numbered steps in logical order
• Clear decision points (“If X, then Y”)
• Checklists for complex multi-step processes
• Specific actions, not vague guidance

5. Examples Quality (20%)

Does the skill show concrete input/output pairs?

This is the most heavily weighted criterion. Examples must be concrete, not abstract. Each example should show:

• A specific input
• The process applied
• The specific output

Critical deduction: Abstract examples instead of concrete I/O pairs = -10 points.

6. Completeness (20%)

Does the skill cover edge cases and provide defaults?

This criterion checks for:

• Edge case handling
• Common pitfalls section
• Templates or frameworks where applicable
• Defaults rather than many options (reduce decision fatigue)

Score Distribution

Note

Grading is deliberately harsh. Here’s the expected distribution:

Range	Frequency	What It Takes
80-100	Rare (~5%)	Exceptional: concrete examples, perfect format, actionable throughout
60-79	Common (~30%)	Good: solid methodology with some gaps in examples or completeness
40-59	Most common (~45%)	Average: has process but lacks concrete examples or over-explains
0-39	Some (~20%)	Poor: vague input, placeholder content, no real methodology

Critical Deductions Summary

Issue	Points Lost
First/second person in description	-10
No Quick Start section	-15
Abstract examples (not input/output pairs)	-10
Over-explaining basics the AI knows	-10
Placeholder content, no real methodology	Scores 0-20 (F)

How to Maximize Your Score

1. Provide detailed input - The generator can only work with what you give it
2. Include real examples - Walk through actual cases with specific inputs and outputs
3. Name your tools - “LinkedIn Recruiter” scores better than “recruiting tools”
4. Describe your process - Sequential steps with decision criteria
5. Answer extraction questions - If prompted, take the time to answer thoroughly