compact
Pass
Audited by Gen Agent Trust Hub on Apr 25, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill uses redundant and forceful imperative language ('MANDATORY IMPORTANT MUST ATTENTION') multiple times throughout the document to prioritize its instructions over the agent's default operational logic.
- [PROMPT_INJECTION]: The skill mandates the use of the 'TaskCreate' tool for all work components and every file read, which attempts to override the agent's autonomous task planning and native decision-making capabilities.
- [PROMPT_INJECTION]: The skill imposes artificial constraints on the agent's reasoning, such as requiring 'confidence >80% to act' and specific evidence patterns ('file:line'), which are attempts to programmatically restrict the agent's model through natural language instructions.
Audit Metadata