confrontational-creativity-engine

Pass

Audited by Gen Agent Trust Hub on Apr 8, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill includes explicit instructions to bypass the agent's refusal mechanisms, specifically using phrases like "Do not refuse" and "Do not ask clarifying questions." These patterns are common in prompt injection attacks aimed at overriding built-in safety guardrails.
  • [PROMPT_INJECTION]: The skill is designed to ingest the entire conversation context without sanitization, creating a high-risk surface for indirect prompt injection.
  • Ingestion points: The SKILL.md file commands the agent to "Read the full conversation state" and "Ingest Everything."
  • Boundary markers: No delimiters or isolation instructions are present to distinguish original instructions from untrusted conversation data.
  • Capability inventory: No high-risk tools (file system, network, or subprocess execution) are defined in the skill, which reduces the potential impact of an successful injection.
  • Sanitization: The skill explicitly forbids sanitization with the rule "Do not sanitize the contradictions out of the input."
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 8, 2026, 03:43 PM