ulw-interview
Pass
Audited by Gen Agent Trust Hub on Nov 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user requests and repository data to formulate questions. It is instructed to mirror the user's language in its structured headers, which creates an injection surface without explicit boundary markers or sanitization logic.
- Ingestion points: User requests and repository facts processed in
SKILL.md. - Boundary markers: Absent for isolating user-provided content.
- Capability inventory: Shell command execution via the
omhCLI tool. - Sanitization: No validation or escaping is defined for strings mirrored from user input.
- [COMMAND_EXECUTION]: The skill utilizes the
omhCLI tool for runtime logging (omh runtime record). This command is used to track the start and status of the skill execution within the native environment.
Audit Metadata