JHJae Hugh AI Lab
Skill Safety Scanner

Inspect before you install

Know what an agent skill can touch.

Review prompts, manifests, and scripts for risky execution, secret access, broad permissions, obfuscation, and network behavior.

Local scanning by default. Nothing is executed.

Skill or code0 characters
Risk reportSafety score --/100
0critical
0high
0medium
0findings
Paste or open a skill, then scan it. Findings explain the observed pattern and the review action it deserves.
Awaiting scan