Tp4
- Category
- MCP Tool Poisoning
- Confidence
- 96% confidence
- Finding
The skill markets itself as an offline delegation, continuation, and acceptance workflow tied to AI-HIVE and authorized tools, but the file mainly contains procedural text and references rather than enforceable orchestration, continuation, or verification logic. This mismatch can mislead users or downstream agents into believing external actions, task handoff, or quality validation occurred when they did not, creating unsafe reliance and possible unauthorized or incomplete task handling.
- Content
