ai-partner-chat
Fail
Audited by Socket on May 4, 2026
1 alert found:
MalwareMalwareSKILL.md
HIGHMalwareHIGH
SKILL.md
SUSPICIOUS. The core note-indexing and local vector DB behavior matches the stated purpose, and the visible install path is standard pip-based tooling. However, the skill's defining feature—having the agent generate and execute code after analyzing arbitrary note contents—creates a disproportionate indirect prompt-injection/execution risk for a chat-memory skill, even without clear evidence of exfiltration or credential theft.
Confidence: 83%Severity: 59%
Audit Metadata