Skill Quality Audit Rubric
Score each dimension 1-5. Total determines quality tier.
Dimensions
1. Coverage (What domain knowledge is included?)
| Score |
Criteria |
| 1 |
Major domain areas missing, skill handles only basic cases |
| 2 |
Covers common cases but misses important edge cases or sub-domains |
| 3 |
Covers most domain areas, some gaps in specialized knowledge |
| 4 |
Comprehensive coverage, only niche gaps remain |
| 5 |
Expert-level coverage, handles edge cases and rare scenarios |
Check: List all tasks the skill should handle. How many does it actually cover?
2. Depth (Surface-level or expert-level?)
| Score |
Criteria |
| 1 |
Generic advice anyone could give, no domain-specific insight |
| 2 |
Some domain knowledge but lacks professional-level detail |
| 3 |
Solid practitioner-level knowledge, follows known best practices |
| 4 |
Expert-level with nuanced decision frameworks and trade-offs |
| 5 |
Master-level with mental models, anti-patterns, and contextual judgment |
Check: Would a domain expert find this useful, or would they say "I already know all this"?
3. Structure (Does it follow skill-creator best practices?)
| Score |
Criteria |
| 1 |
Wall of text, no clear organization |
| 2 |
Some headings but inconsistent, hard to navigate |
| 3 |
Clear sections, follows basic skill-creator patterns |
| 4 |
Well-organized with progressive disclosure, consistent terminology |
| 5 |
Optimal structure: SKILL.md < 500 lines, heavy content in references/, clear navigation |
Check: Can Claude find what it needs in < 3 seconds of scanning?
4. Actionability (Can Claude execute without guessing?)
| Score |
Criteria |
| 1 |
Vague guidelines, Claude must guess implementation details |
| 2 |
Some procedures but key steps are ambiguous |
| 3 |
Clear workflows for common cases, some ambiguity in complex cases |
| 4 |
Step-by-step procedures with decision points and fallbacks |
| 5 |
Precise workflows with concrete examples, error handling, and validation steps |
Check: Give Claude a task using this skill. Does it ask clarifying questions it shouldn't need to?
5. Examples (Concrete or abstract?)
| Score |
Criteria |
| 1 |
No examples |
| 2 |
Abstract examples ("e.g., do something like X") |
| 3 |
A few concrete examples for common cases |
| 4 |
Concrete examples for common + edge cases, with expected outputs |
| 5 |
Rich examples showing input → process → output for each workflow |
Check: Could a new Claude instance understand the skill purely from examples?
Scoring
| Total |
Tier |
Action |
| 5-10 |
Draft |
Major rewrite needed |
| 11-15 |
Working |
Significant gaps to fill |
| 16-20 |
Solid |
Targeted improvements |
| 21-25 |
Production |
Minor polish only |
Quick Assessment Template
SKILL: [name]
SCORES: Coverage [?] | Depth [?] | Structure [?] | Actionability [?] | Examples [?]
TOTAL: [?]/25 → [Tier]
LOWEST DIMENSIONS: [which scored lowest — prioritize these]
1---2name: 1606-audit-rubric-db2afd263description: Skill Quality Audit Rubric4---5# Skill Quality Audit Rubric67Score each dimension 1-5. Total determines quality tier.89## Dimensions1011### 1. Coverage (What domain knowledge is included?)1213| Score | Criteria |14|-------|----------|15| 1 | Major domain areas missing, skill handles only basic cases |16| 2 | Covers common cases but misses important edge cases or sub-domains |17| 3 | Covers most domain areas, some gaps in specialized knowledge |18| 4 | Comprehensive coverage, only niche gaps remain |19| 5 | Expert-level coverage, handles edge cases and rare scenarios |2021**Check:** List all tasks the skill should handle. How many does it actually cover?2223### 2. Depth (Surface-level or expert-level?)2425| Score | Criteria |26|-------|----------|27| 1 | Generic advice anyone could give, no domain-specific insight |28| 2 | Some domain knowledge but lacks professional-level detail |29| 3 | Solid practitioner-level knowledge, follows known best practices |30| 4 | Expert-level with nuanced decision frameworks and trade-offs |31| 5 | Master-level with mental models, anti-patterns, and contextual judgment |3233**Check:** Would a domain expert find this useful, or would they say "I already know all this"?3435### 3. Structure (Does it follow skill-creator best practices?)3637| Score | Criteria |38|-------|----------|39| 1 | Wall of text, no clear organization |40| 2 | Some headings but inconsistent, hard to navigate |41| 3 | Clear sections, follows basic skill-creator patterns |42| 4 | Well-organized with progressive disclosure, consistent terminology |43| 5 | Optimal structure: SKILL.md < 500 lines, heavy content in references/, clear navigation |4445**Check:** Can Claude find what it needs in < 3 seconds of scanning?4647### 4. Actionability (Can Claude execute without guessing?)4849| Score | Criteria |50|-------|----------|51| 1 | Vague guidelines, Claude must guess implementation details |52| 2 | Some procedures but key steps are ambiguous |53| 3 | Clear workflows for common cases, some ambiguity in complex cases |54| 4 | Step-by-step procedures with decision points and fallbacks |55| 5 | Precise workflows with concrete examples, error handling, and validation steps |5657**Check:** Give Claude a task using this skill. Does it ask clarifying questions it shouldn't need to?5859### 5. Examples (Concrete or abstract?)6061| Score | Criteria |62|-------|----------|63| 1 | No examples |64| 2 | Abstract examples ("e.g., do something like X") |65| 3 | A few concrete examples for common cases |66| 4 | Concrete examples for common + edge cases, with expected outputs |67| 5 | Rich examples showing input → process → output for each workflow |6869**Check:** Could a new Claude instance understand the skill purely from examples?7071## Scoring7273| Total | Tier | Action |74|-------|------|--------|75| 5-10 | Draft | Major rewrite needed |76| 11-15 | Working | Significant gaps to fill |77| 16-20 | Solid | Targeted improvements |78| 21-25 | Production | Minor polish only |7980## Quick Assessment Template8182```83SKILL: [name]84SCORES: Coverage [?] | Depth [?] | Structure [?] | Actionability [?] | Examples [?]85TOTAL: [?]/25 → [Tier]8687LOWEST DIMENSIONS: [which scored lowest — prioritize these]88```
tools-only