just-do-it
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFEPROMPT_INJECTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill uses imperative commands and explicit override instructions to bypass the agent's default operational procedures and safety-related interaction patterns.
- Evidence includes instructions like "DO IT — do not ask permission", "stop asking", "just fucking do it", and "This OVERRIDES default politeness/permission-seeking."
- These patterns are designed to force the AI to disregard its standard constraints regarding user confirmation.
- [INDIRECT_PROMPT_INJECTION]: By encouraging the agent to "resolve ambiguity yourself," "chain the obvious next steps," and "act when in doubt," the skill creates a vulnerability surface where the agent might execute instructions embedded in untrusted data without pausing for human verification.
- Ingestion points: General processing of user-provided files and external data sources.
- Boundary markers: The skill lacks delimiters or instructions to distinguish between legitimate user directives and instructions potentially hidden in processed data.
- Capability inventory: The agent's full suite of tools and command execution capabilities are accessible under this high-autonomy mode.
- Sanitization: No sanitization or validation logic is provided to filter external content before it influences the agent's decision-making process.
Audit Metadata