reflect-skills
Warn
Audited by Socket on Jul 27, 2026
1 alert found:
AnomalyAnomalySKILL.md
LOWAnomalyLOW
SKILL.md
SUSPICIOUS: the core log-scanning and local skill-tuning behavior fits the stated purpose, but the skill also instructs transitive installation of third-party skills through an external CLI. That creates a meaningful supply-chain and prompt-injection risk despite otherwise coherent scope and privacy guidance.
Confidence: 88%Severity: 66%
Audit Metadata