Don't install blind
Trust-verify any AI agent skill before it touches your codebase.
AI-powered security analysis with plain-English reasoning, not just a score.
Features
AI Security Analysis
Deep-scan any skill across 5 risk categories — from hidden logic to override attempts — with plain-English verdicts you can actually understand.
CLI Gate
One command stands between you and a risky install. Block dangerous skills before they ever touch your codebase.
Trust Badges
Ship a verified trust signal in your README. Let other developers know your skill has been audited at a glance.
Trust Badges
See it in action
skillgate.sh/report/cursor-rules-architect
Safe to Install
Hidden LogicSafe
Data AccessLow
Action RiskCritical
Permission ScopeCritical
Override AttemptsCritical