ok-i-go-with-your-recommendations
Pass
Audited by Gen Agent Trust Hub on Jul 22, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill contains instructions to bypass standard human-in-the-loop (HITL) confirmation cycles. By specifying "Do not re-pitch options or ask for confirmation again," the skill overrides a key security control intended to prevent autonomous agents from performing unintended or high-impact actions without explicit user approval.
- [PROMPT_INJECTION]: The skill exposes the agent to indirect prompt injection risks by automating the execution of prior conversation state. It lacks boundary markers or validation steps to ensure that the "prior recommendations" being executed were not influenced by malicious instructions embedded in external data (such as files, websites, or emails) processed earlier in the session.
- [PROMPT_INJECTION]: The instruction to "pick the clearest prior plan, state the one assumption, then proceed" when plans are ambiguous allows the agent to make autonomous decisions and execute them without verifying the assumption with the user, further reducing human oversight.
Audit Metadata