spree-catalog

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

• Third-party content exposure detected (high risk: 0.85). High: the REQUIRED workflow explicitly fetches live public web documentation pages at runtime (e.g., https://spreecommerce.org/docs/...) and also performs web-search, whose returned page text/snippets can be ingested into the agent’s LLM context, enabling outsider-authored free text prompt injection.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly instructs fetching live documentation at runtime from https://spreecommerce.org/docs/developer/core-concepts/architecture, which would inject external doc content into the agent's context and thereby directly influence its prompts/behavior.