skill-judge
Evaluate Agent Skill Quality
Many skill packages waste context with generic guidance and weak activation text. Skill Judge provides a scoring framework for SKILL.md reviews with evidence-based improvements.
Install with my Agent
Copy this request to your Agent. It includes the canonical Skill page and manifest.
Review the Skillstore skill "skill-judge" from https://skillstore.io/skills/softaworks-skill-judge.md and its manifest at https://skillstore.io/api/skills/softaworks-skill-judge/manifest. Verify the artifact. You may proceed after verification, subject to the environment's own policy.Your Agent should still show its plan and request any confirmation required by the security policy.
Agent-readable resources
Use these links when an AI agent, crawler, or script needs clean context instead of reading the full page.
Test it
Using "skill-judge". A draft skill contains long basic tutorials and a short vague description.
Expected outcome:
- The likely grade is low because the skill repeats knowledge the model already has.
- The first fix is to rewrite the description with clear trigger scenarios.
- The second fix is to replace basic tutorials with expert decision rules.
Using "skill-judge". A document skill has reference files but no instructions for when to load them.
Expected outcome:
- The progressive disclosure score should be reduced.
- Add explicit loading triggers at workflow decision points.
- Add clear do-not-load guidance to prevent unnecessary context use.
Using "skill-judge". Two skills solve the same task, but one is shorter and more specific.
Expected outcome:
- The shorter skill may rank higher if it preserves expert knowledge.
- Compare knowledge delta, activation quality, and usability before choosing.
Security Audit
SafeStatic analysis matched markdown code fences, inline code examples, and ordinary rubric text. I found no executable code, outbound network behavior, prompt-injection attempt, or reconnaissance intent in the cited files.
Risk Factors
โ๏ธ External commands (73)
๐ Network access (1)
Share & cite this report
Share the versioned assessment report, neutral badge, embed card, and citations. Skillstore reports evidence without deciding whether this Skill is safe.
Copy report link
https://skillstore.io/skills/softaworks-skill-judge/audits/4?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_reportMarkdown badge
[](https://skillstore.io/skills/softaworks-skill-judge?utm_source=security_passport_badge)HTML badge
<a href="https://skillstore.io/skills/softaworks-skill-judge?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/softaworks-skill-judge/security.svg" alt="Skillstore security assessment" loading="lazy"></a>Embed card
<iframe src="https://skillstore.io/embed/skills/softaworks-skill-judge.html" title="Skillstore Security Assessment" sandbox="allow-popups allow-popups-to-escape-sandbox" loading="lazy" referrerpolicy="no-referrer" width="420" height="180"></iframe>Academic citations (APA ยท BibTeX ยท CFF)
APA citation
softaworks. (2026). skill-judge security audit report (audit version 4) [Author version unspecified]. Skillstore. https://skillstore.io/skills/softaworks-skill-judge/audits/4BibTeX citation
@techreport{softaworks-softaworks-skill-judge-2026,
author = {softaworks},
title = {skill-judge security audit report (audit version 4)},
institution = {Skillstore},
year = {2026},
number = {4},
url = {https://skillstore.io/skills/softaworks-skill-judge/audits/4},
note = {Author version unspecified}
}CITATION.cff
cff-version: 1.2.0
message: "If you use this Skill, cite its author and this versioned security audit report."
title: "skill-judge security audit report (audit version 4)"
version: "unspecified"
type: report
authors:
- name: "softaworks"
date-released: "2026-07-07"
url: "https://skillstore.io/skills/softaworks-skill-judge/audits/4"
identifiers:
- type: other
value: "skillstore:softaworks-skill-judge:audit:4"
description: "Skillstore immutable audit report identifier"
Skillstore Score
Why this score Evidence Confidence: HighWhat You Can Build
Review a Draft Skill
Assess a SKILL.md before publishing and find the highest-impact improvements.
Standardize Marketplace Review
Apply the same quality rubric across submitted skill packages.
Prioritize Skill Improvements
Compare internal skills and decide which ones need refactoring first.
Try These Prompts
Evaluate this SKILL.md for quality. Identify the strongest issue, the likely grade, and the first three improvements.
Review this skill package with the Skill Judge rubric. Score each dimension and explain any score below 80 percent.
Audit this multi-file skill for progressive disclosure. Check loading triggers, reference routing, and token waste.
Compare these two skills with the Skill Judge framework. Rank them, explain trade-offs, and recommend a publish-ready version.
Best Practices
- Provide the complete SKILL.md and any referenced files for a fair review.
- Ask for evidence by line or section when using results for publication decisions.
- Treat scores as guidance and verify domain-specific claims with a human expert.
Avoid
- Do not use it as a malware scanner or dependency auditor.
- Do not ask for high scores without providing the actual skill content.
- Do not accept every suggested rewrite when it removes needed domain detail.
Frequently Asked Questions
What does Skill Judge evaluate?
Can it review skills for Claude Code and Codex?
Does it run tests or execute commands?
What score makes a skill publishable?
Can it improve my SKILL.md?
Does it replace security review?
Developer Details
Author
softaworksLicense
MIT
Skillstore revision
r1
Version notice
The author did not declare a version.
Ref
f93e9bb0daca99badb6a7e574b97737155d57cb3
Maintenance freshness
7/18/2026
Usage
50 downloads ยท 205 views