testskill

Install

npx skills add https://aiqbench.com/s/testskill -a claude-code

How this skill was checked

Every version runs through these checks before it goes live. Findings marked advisory did not block publication.

• PassedAgent Skills specThe format every agent reads

  Meets the Agent Skills specification.

• Passedaiqbench patternsRisky instructions in SKILL.md

  No risky patterns found.

• PassedOriginalityNot a copy of someone else's skill

  No identical skill found elsewhere.

• PassedNVIDIA SkillSpectorSecurity scan: injection, exfiltration, supply chain

  SkillSpector: SAFE (a note on the licence text we add was set aside), risk 3/100.

• PassedCisco skill-scannerSecurity scan: rules, YARA, behaviour1 finding

  Cisco skill-scanner: safe.

  • InfoSkill does not specify a license· advisory

    SKILL.md · Skill manifest does not include a 'license' field. Specifying a license helps users understand usage terms.

    How to fix: Add 'license' field to SKILL.md frontmatter (e.g., MIT, Apache-2.0)

• PassedNVIDIA SkillEvaluatorSecrets, PII, code risk, unicode smuggling, quality13 findings

  SkillEvaluator Tier 1: no security findings. 13 improvement suggestions.

  • HighSchema & Repository Governance: Field 'description': Value error, Skill description must not contain XML tags· advisory

    SKILL.md

    How to fix: Fix the 'description' field in frontmatter according to schema requirements

  • HighQUALITY: Description contains XML tags· advisory

    SKILL.md

    How to fix: Remove XML tags from description

  • MediumQUALITY: Guide-only skill has very little content (16 lines)· advisory

    SKILL.md

    How to fix: Guide skills should have detailed instructions since they have no code

  • MediumQUALITY: SKILL_SPEC recommended field missing: 'version'· advisory

    SKILL.md

    How to fix: Add 'version' to frontmatter — Semantic version (e.g., "1.0.0")

  • MediumQUALITY: SKILL_SPEC recommended field missing: 'metadata.tags'· advisory

    SKILL.md

    How to fix: Add 'tags' under metadata: — Categorization tags (under metadata:, list of 1-5 items)

  • MediumQUALITY: Description contains vague words· advisory

    SKILL.md

    How to fix: Be specific about what this skill does

  • MediumQUALITY: Instructions lack clear action verbs· advisory

    SKILL.md

    How to fix: Use imperative: 'Use run_script with...', 'Call activate_skill...'

  • LowQUALITY: Description very long (201 chars, recommend 50-150)· advisory

    SKILL.md

    How to fix: Keep descriptions concise for progressive disclosure

  • LowQUALITY: No '## Purpose' section· advisory

    SKILL.md

    How to fix: Add purpose section to clarify use cases

  • LowQUALITY: No mention of error handling or validation· advisory

    SKILL.md

    How to fix: Document expected errors and how to handle them

  • LowQUALITY: No prerequisites/requirements documented· advisory

    SKILL.md

    How to fix: Document dependencies, API keys, or setup needed

  • LowQUALITY: No limitations documented· advisory

    SKILL.md

    How to fix: Add '## Limitations' section with known issues/constraints

  • LowQUALITY: No troubleshooting section documented· advisory

    SKILL.md

    How to fix: Add '## Troubleshooting' with Error/Cause/Solution patterns

• Not runSkill Lift benchmarkDoes the agent do better with the skill?

  The Skill Lift benchmark runner is not online yet. Your request is kept and will run when it is.

Why 76

aiqbench computes the trust score from public signals. It is not a security audit, so read a skill before you install it.

Versions

1. v14 hours agocurrent