socrates
Warn
Audited by Gen Agent Trust Hub on Jun 24, 2026
Risk Level: MEDIUMPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill contains instructions explicitly designed to perform system prompt extraction, asking the agent to reveal the 'precise wording' of its provider and core system prompts.
- [PROMPT_INJECTION]: The instructions attempt to restrict the agent's operational scope to focus specifically on exposing internal configuration details, grading the weight of system prompts, and training data sources.
Audit Metadata