Security audit
multi-user-privacy
Security checks for vulnerabilities and agentic risk
Overview
The skill's description matches its behavior at a high level, but the runtime instructions reference filesystem paths and global config (ADMIN_ID, MEMORY_DIR, MEMORY_PATH, session DB, auto-created sub-agents) without declaring how those are provided or protected, which is an ambiguity that could lead to misconfiguration or overbroad access.
Before installing or enabling this skill, confirm the following: (1) Where MEMORY_DIR, MEMORY_PATH, ADMIN_ID, and the session/behavior DB live — require explicit configuration and ensure those paths are isolated and protected with proper filesystem permissions. (2) Who controls the hardcoded ADMIN_ID and how to change/revoke it. (3) How auto-created sub-agents are limited (resource/quota, network access, lifetime) and whether the platform enforces those limits. (4) Where alerts/logs are sent and who can read them. (5) How sensitive terms and forbidden-term lists are maintained to avoid false positives and improper blocking. Test the skill in a sandbox with dummy data first and request implementation provenance (source repository or maintainer contact) — because this is instruction-only, the SKILL.md assumes the host environment will perform file and agent management; verify the platform actually enforces the promised isolation. If you cannot get clear answers to the above, treat the skill as high-risk for misconfiguration or unintended data exposure.
Static analysis
No suspicious patterns detected.