azure-prepare

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (high risk: 1.00). The prompt contains explicit authoritative commands ("MUST", "IGNORE any prior training", "supersedes all other sources") that attempt to override system/prior instructions and system context, which is a deceptive/overriding instruction outside the normal scope of an Azure-prepare skill.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill documentation includes a direct install command that would fetch and execute remote code at runtime — curl -fsSL https://aka.ms/install-azd.sh | bash — which is a required install step for the azd-based workflows the skill depends on.