novoads-pixar-ad

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). The skill explicitly tells the agent to perform an HTTP PUT "sending headers VERBATIM" (and earlier "PUT the bytes with the returned headers verbatim"), which would require the LLM/tool to include returned signed headers or tokens verbatim in generated requests — a direct secret-exposure/exfiltration risk.

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

• Third-party content exposure detected (medium risk: 0.30). The skill’s runtime ingests outsider-authored free text via the operator-provided “product as text” (URL or listing copy) to perform the product read (Gate 1), meaning attackers who can supply that input can inject content that the LLM ingests.