Tp4
- Category
- MCP Tool Poisoning
- Confidence
- 99% confidence
- Finding
The skill advertises calendar conflict analysis, scheduling outputs, and optional AI-HIVE/MCP orchestration, but the analyzed content does not substantiate those behaviors and instead points to unrelated/offline validation behavior. This kind of description-behavior mismatch is dangerous because it can mislead users and operators into granting trust, permissions, or sensitive calendar data under false assumptions about what the skill actually does.
- Content
