Security audit
Security checks for vulnerabilities and agentic risk
This is a meeting-assistant skill that is coherent and disclosed, with sensitive calendar/task use limited by draft-first and user-authorization instructions.
Install only if you are comfortable letting the host agent use meeting-related calendar, transcript, attendee, and task context. Review drafts before approving distribution or task syncing, especially where dissent, sensitive decisions, or recurring commitments are included.
Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.
---
name: Meeting
version: 2.0.1
description: "Orchestrate meeting preparation, capture, and commitment tracking with draft-first confirmation"
The trigger phrases are broad enough to match common, nonspecific productivity requests such as preparing for meetings, generating minutes, or tracking commitments. In an agent ecosystem, this can cause unintended invocation or routing collisions, leading the skill to capture user context or influence workflows outside its intended scope.
The documentation makes concrete behavioral claims such as requiring user approval before distribution or tracking and prohibiting autonomous recording. However, this artifact is only a README and contains no implementation, so the stated operational guarantees are unsupported here and could misrepresent actual behavior elsewhere in the skill.
No suspicious patterns detected.