Lp3
Medium
- Category
- MCP Least Privilege
- Confidence
- 82% confidence
- Finding
- The skill invokes shell commands (`curl`, `jq`) and handles sensitive authentication material (`APP_KEY`), but it declares no permissions or equivalent capability boundaries. This creates a real security issue because consumers may not understand that the skill performs network access and shell execution, increasing the risk of unexpected outbound requests, credential exposure, or misuse in environments that rely on permission declarations for policy enforcement.
