quality-grading — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited quality-grading (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Evaluate and improve code, specifications, or design documents across four quality dimensions with calibrated scoring and auto-fix capabilities.
This skill provides a multi-dimensional quality grading system for any artifact type:
It supports two invocation modes:
The skill targets score 4+ as the quality bar. Auto-fix attempts are limited to 2 per dimension to keep improvements pragmatic and avoid unnecessary complexity.
Load this skill when:
artifact: Path to the file, directory, or snippet to grade (required)mode: evaluate or grade-and-fix (default: evaluate)Measures: Architecture clarity, scalability, and separation of concerns
For Code: Module structure, dependency management, separation of concerns For Designs: Architecture clarity, module boundaries, scalability patterns, diagram quality For Specifications: Logical organization, traceability, clarity of structure
What to look for:
Scoring Rubric:
| Score | Descriptor |
|---|---|
| 5 | Exceptional architecture with clear module boundaries, proven scalability patterns, strong separation of concerns |
| 4 | Good architecture with clear modules, adequate scalability consideration, minor separation issues |
| 3 | Adequate architecture with recognizable modules, moderate separation of concerns |
| 2 | Weak architecture with unclear module boundaries, limited separation of concerns |
| 1 | Poor architecture with tangled dependencies, no clear modules, intermingled concerns |
Auto-Fix (score < 4):
Measures: Avoidance of generic boilerplate and tailoring to the problem
For Code: Problem-specific solutions vs copy-paste patterns For Designs: Domain-specific architecture vs generic templates For Specifications: Tailored requirements vs boilerplate language
What to look for:
Scoring Rubric:
| Score | Descriptor |
|---|---|
| 5 | Highly tailored solution with novel approaches directly mapped to the problem domain |
| 4 | Mostly tailored solution with some unique adaptations to the problem |
| 3 | Mix of generic and tailored approaches |
| 2 | Mostly generic with minimal adaptation to the problem |
| 1 | Generic, copy-paste solution with no adaptation to the specific problem |
Auto-Fix (score < 4):
Measures: Code cleanliness, error handling, and documentation
For Code: Clean code, error handling, comments, documentation For Designs: Diagram clarity, consistent formatting, completeness For Specifications: Clear writing, consistent terminology, proper structure
What to look for:
Scoring Rubric:
| Score | Descriptor |
|---|---|
| 5 | Exceptionally clean code with comprehensive error handling and exemplary documentation |
| 4 | Clean code with good error handling and documentation with minor gaps |
| 3 | Adequate code quality with some error handling and documentation |
| 2 | Messy code with limited error handling and minimal documentation |
| 1 | Messy, error-prone code with little to no documentation |
Auto-Fix (score < 4):
Measures: Feature completeness and edge case handling
For Code: Features work, edge cases handled For Designs: All requirements covered, feasibility, completeness For Specifications: Complete requirements, clear acceptance criteria, no gaps
What to look for:
Scoring Rubric:
| Score | Descriptor |
|---|---|
| 5 | Complete feature implementation with comprehensive edge case handling |
| 4 | Full implementation with minor edge case gaps |
| 3 | Most features implemented with some edge cases handled |
| 2 | Partial implementation with several unhandled edge cases |
| 1 | Missing features and unhandled edge cases causing frequent failures |
Auto-Fix (score < 4):
When invoked in grade-and-fix mode:
Use the examples in references/ directories to calibrate your scoring:
Compare the artifact you're grading against these examples for consistent calibration.
The skill produces JSON output:
{
"artifact": "<path-to-graded-file>",
"artifactType": "code | specification | design",
"timestamp": "<ISO-8601 timestamp>",
"mode": "evaluate | grade-and-fix",
"dimensions": {
"designQuality": {
"score": 1-5,
"justification": "<brief explanation of the score>",
"improved": "<null | description of auto-fix applied>"
},
"originality": {
"score": 1-5,
"justification": "<brief explanation of the score>",
"improved": "<null | description of auto-fix applied>"
},
"craft": {
"score": 1-5,
"justification": "<brief explanation of the score>",
"improved": "<null | description of auto-fix applied>"
},
"functionality": {
"score": 1-5,
"justification": "<brief explanation of the score>",
"improved": "<null | description of auto-fix applied>"
}
},
"overallScore": "<average of dimensions, displayed as X.Y>",
"summary": "<overall assessment text>",
"suggestions": {
"<dimension>": "<actionable improvement if score < 4 and auto-fix failed>"
}
}| Term | Definition |
|---|---|
| Quality Grading Skill | A skill that evaluates artifacts across multiple dimensions using a 1-5 scoring scale |
| Dimension | A distinct aspect of quality being evaluated (Design Quality, Originality, Craft, Functionality) |
| Artifact Type | The category of artifact being graded: code, specification, or design |
| Few-Shot Calibration | Providing example artifacts with known scores to guide consistent grading |
| Self-Healing | The skill's ability to auto-fix quality gaps without user prompting |
| Grade-and-Fix Mode | Invocation mode where grading and improvement are performed sequentially without user input |
User: Grade this implementation using quality-grading skill
Artifact: src/services/auth.ts
Mode: evaluateUser: Assess the quality of this architecture design
Artifact: docs/architecture/system-design.md
Mode: grade-and-fixUser: Grade this requirements document
Artifact: specs/changes/feature/requirements.md
Mode: evaluateUser: Grade and improve this module
Artifact: src/utils/
Mode: grade-and-fix~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.