sandbase
Warn
Audited by Socket on Aug 19, 2026
1 alert found:
AnomalyAnomalySKILL.md
LOWAnomalyLOW
SKILL.md
SUSPICIOUS: the skill is internally coherent as a unified AI/API gateway, but it centralizes a very broad set of external capabilities and data flows through SandBase rather than official per-provider APIs. No clear malware indicators or malicious exfiltration instructions are present, but the intermediary routing and underdocumented `sandbase connect` trust path make it a medium-risk skill.
Confidence: 82%Severity: 62%
Audit Metadata