auto
Pass
Audited by Gen Agent Trust Hub on May 11, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill allows external data to influence the agent's behavior via the goal parameter.\n
- Ingestion points: The
goalargument passed toouroboros_autoinSKILL.mdis a direct entry point for untrusted instructions.\n - Boundary markers: None. There are no instructions to isolate the goal text or treat it as data only.\n
- Capability inventory: The tool
ouroboros_autocan generate, modify, and execute files on the system.\n - Sanitization: None. The skill does not perform any validation on the provided goal before processing it.\n- [COMMAND_EXECUTION]: The skill is designed to run generated code automatically once it reaches a certain quality threshold. This is a high-risk capability inherent to autonomous development tools.
Audit Metadata