gcp-diagram
Warn
Audited by Socket on May 6, 2026
1 alert found:
AnomalyAnomalySKILL.md
LOWAnomalyLOW
SKILL.md
SUSPICIOUS. The core purpose is coherent and the direct Vertex AI path is benign, but the skill explicitly prioritizes an unofficial third-party MCP tool before the official Google path. That creates moderate supply-chain and data-flow uncertainty, especially if sensitive architecture details or credentials are routed through PaperBanana or a custom gateway. No clear malware or overt exfiltration is shown, but the third-party-first design makes the overall skill riskier than a pure Vertex AI implementation.
Confidence: 84%Severity: 52%
Audit Metadata