Security audit
Security checks for vulnerabilities and agentic risk
This skill is a disclosed late-night emotional check-in prompt with no executable code, but users should be comfortable with its proactive one-message behavior.
Install this only if you want a skill that may proactively open a late-night emotional check-in when you are active or online during the configured quiet-hours window. Configure DND or opt-out preferences, and avoid enabling memory for sensitive emotional details unless you are comfortable with the host product retaining them.
The skill description explicitly says it activates automatically during late-night hours without a manual trigger. In an emotional-support context, broad proactive activation can cause unsolicited outreach based on weak signals such as mere online presence, creating privacy, consent, and boundary risks if users did not clearly opt in to nighttime check-ins.
The auto-trigger logic is broad and underspecified: it fires when local time is in a wide window and either the user was active after 22:00 or the session is online, with only a narrow DND exclusion. Because 'session is online' is vague and negative constraints are minimal, the assistant could enter an emotional-support mode and proactively message users who did not expect or want intervention, especially in sensitive circumstances.
No suspicious patterns detected.
