active-record-encryption
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCREDENTIALS_UNSAFECOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill has a significant attack surface for indirect prompt injection because it ingests user-supplied information to generate and execute Ruby code via administrative tools.
- Ingestion points: Sensitive field names, model structures, and configuration requirements provided by the user in the conversation.
- Boundary markers: None present in the instructions to separate user data from generated code logic.
- Capability inventory: The skill uses
Bash(bin/rails runner),Bash(bin/rails console), andWrite/Edittools, allowing for arbitrary Ruby code execution and file system modifications. - Sanitization: No explicit sanitization or validation steps are defined for data interpolated into shell commands or generated Ruby files.
- [CREDENTIALS_UNSAFE]: The documentation provides specific, literal strings as example encryption keys (e.g.,
primary_key: YehXdfzxVKpoLvKseJMJIEGs2JxerkB8). While intended as placeholders, these are valid-looking keys that a user might copy-paste into production configuration instead of generating unique, random values as intended. - [COMMAND_EXECUTION]: The skill relies on powerful Rails administrative commands (
bin/rails runner,bin/rails console) to verify configurations and perform data migrations. These tools allow for unrestricted code execution within the application environment.
Audit Metadata