Tp4
High
- Category
- MCP Tool Poisoning
- Confidence
- 95% confidence
- Finding
- The skill's documented behavior does not match the reported implementation: it allegedly uses an undocumented API, exposes extra data-processing capabilities, and returns placeholder/mock values while claiming to fetch real city-detail data. This is dangerous because users and downstream agents may make decisions based on incorrect assumptions about data provenance, authenticity, and completeness, which can enable silent data integrity failures and reduce oversight of undocumented network behavior.
