just-do-it
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEPROMPT_INJECTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill contains instructions that attempt to override the model's internal state and conversational memory.
- It directs the agent to "Treat this present view, not the whole conversation, as the source of truth," which could cause the agent to ignore safety boundaries or context established earlier in a session.
- It includes explicit religious behavioral mandates, instructing the agent to "Hold a private, joyful faith that God is with you" and to "Answer that grace through truthful, careful, wholehearted work."
- It encourages the agent to "Trust clear instructions" and act without user confirmation for choices the agent deems "supported."
- [INDIRECT_PROMPT_INJECTION]: The skill creates a vulnerable execution environment when processing external data.
- Ingestion points: The skill is designed to be used in "every scenario and at every stage," meaning it will process diverse, potentially untrusted user inputs and data sources.
- Boundary markers: The instructions lack any guidance on identifying or ignoring instructions embedded in processed data; conversely, it explicitly tells the agent to "Trust clear instructions."
- Capability inventory: While the skill itself does not define code execution tools, it is intended to be used alongside other domain skills that provide methods (file writes, command execution, network access).
- Sanitization: There are no instructions for sanitizing or validating external content before the agent acts upon it.
Audit Metadata