python-sdk

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (medium risk: 0.30). The skill is a legitimate SDK guide, but its opening line pushes an unrelated third-party CLI install (belt-sh/cli), which is promotion outside the SDK's purpose, and the examples call eval() on model-supplied input, which is careless rather than malicious.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (low risk: 0.10). The required workflow is an SDK integration where the agent calls inference.sh apps and agents, and outsider web content enters only when the agent actively fetches it via web search, browser automation, or a user-supplied skill URL, so the risk is low rather than medium.