Skill Test
PassAudited by ClawScan on May 1, 2026.
Overview
This instruction-only skill is coherent and safety-oriented, with only expected notes around using sub-agents and temporary command-line sandbox installs for testing other skills.
This skill appears safe to install as an instruction-only testing helper. When using it, test with sample data, verify any npx install slug and temporary directory before running commands, and only save skill preference notes if you are comfortable reusing that context later.
Findings (3)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Test skill content and prompts may be sent to a separate agent session for evaluation.
The skill passes skill content and test tasks to sub-agents as part of its isolation strategy. This is expected for the stated purpose, but it is still an inter-agent data flow involving potentially untrusted skill text.
sessions_spawn( task="You have ONE skill loaded: [skill content]. Test by doing: [task]"
Use non-sensitive sample tasks and keep the sub-agent isolated, as the skill recommends.
If followed, the skill can create and remove temporary local directories while testing candidate skills.
The documentation includes local command-line installation and deletion examples. They are scoped to temporary test folders and align with the skill's sandboxing purpose.
npx clawhub install <slug> --dir /tmp/skill-test/<slug> ... rm -rf /tmp/skill-test/<slug>
Verify the slug and target directory before running the commands, and keep cleanup limited to the intended /tmp test folder.
Your skill preferences and task context could influence future recommendations if recorded.
The skill suggests recording evaluation preferences for later use. This is purpose-aligned, but users should be aware if such notes are saved into persistent memory or shared context.
Track for future recommendations: - Which skill won - Why (user's stated reason) - Task context
Only store preference notes when useful, and avoid including sensitive task details in long-term memory.