coderpad

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

• Third-party content exposure detected (high risk: 0.90). This skill explicitly fetches CoderPad user-generated content (e.g., via Membrane actions like get-pad, get-pad-environment, get-pad-events, and list-organization-questions) as shown in the "Popular actions" and "Get Pad Environment" sections of SKILL.md, meaning the agent will read untrusted third-party content that could influence its behavior.