until-done
Pass
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill orchestrates an autonomous Plan-Act-Observe-Verify loop driven by external, potentially untrusted inputs.
- Ingestion points: Ingests user-supplied goals via arguments and retrieves verification commands from local files such as
.agent/context/loops.mdordocs/agents/loops.md(found in SKILL.md). - Boundary markers: The instructions do not define boundary markers or clear separation between the agent's internal logic and the user-supplied goal content.
- Capability inventory: The 'Act' and 'Verify' steps involve executing tool calls, file operations, and shell commands as part of the loop discipline.
- Sanitization: There is no evidence of validation or sanitization for the goal strings or the verification commands extracted from the documentation files.
Audit Metadata