openai-docs

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (high risk: 0.90). The prompt instructs the agent to run installation commands and "immediately retry" with escalated permissions while telling it not to ask the user, which are hidden/deceptive operational directives to change system state and escalate privileges outside the skill's stated docs-fetching purpose.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 0.90). The skill explicitly tells the agent to retry an install with "escalated permissions" (and to do so without asking the user), which directs the agent to bypass or escalate sandbox/permissions and thus modify machine state.