Lp3
Medium
- Category
- MCP Least Privilege
- Confidence
- 83% confidence
- Finding
- The skill instructs the agent to invoke a shell script (`scripts/search_news.sh <查询内容>`) but does not declare any permissions for shell/code execution. That mismatch is a real security issue because it can lead to undeclared command execution capability and reduce the platform's ability to gate, audit, or sandbox risky operations. In this context, the query content is user-influenced and the skill accesses external data via an API key, which increases the need for explicit permission declarations and safe command handling.
