cross-cr
Warn
Audited by Socket on Oct 30, 2026
1 alert found:
SecuritySecuritySKILL.md
MEDIUMSecurityMEDIUM
SKILL.md
SUSPICIOUS: the skill is coherent for cross-review, but it deliberately exports repository content to external model vendors and can forward credentials via --pass-env. The largest additional concern is Antigravity: the skill admits agy may run commands allowed in the user's global settings and loads repo instructions, so the claimed read-only review boundary is weaker than for Codex/Claude. The credential-file findings are mostly exclusion lists, not secret theft, and the backtick finding is benign documentation.
Confidence: 90%Severity: 76%
Audit Metadata