Lp3
Medium
- Category
- MCP Least Privilege
- Confidence
- 90% confidence
- Finding
- The skill invokes shell scripts and external binaries (`bash`, `curl`, `jq`) but does not declare explicit permissions for shell/code execution. That mismatch can weaken platform trust boundaries because an agent or reviewer may treat the skill as lower-risk than it actually is, while it still has the ability to make outbound requests and process untrusted input through the shell. In this context the danger is elevated because the skill is specifically a web-search tool that sends user-supplied queries to a third-party API, creating both command-execution and data-exfiltration exposure if the runtime permission model is not transparent.
