Missing User Warnings
Medium
- Confidence
- 96% confidence
- Finding
- The skill reads up to 20 files from the provided path and prints the first 3000 characters of each, which can expose proprietary source code, embedded secrets, credentials, or sensitive data to the model or downstream logs. Because there is no explicit warning, consent checkpoint, or filtering for sensitive files, a user may unintentionally disclose confidential content simply by pointing the skill at a project directory.
