implementing-plans

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

• Third-party content exposure detected (high risk: 0.80). The skill's "Handle Failures" workflow explicitly instructs using a Task tool with subagent_type "web-researcher" to look up error messages/issues on the web (external libraries/context), which means the agent will fetch and interpret untrusted third-party web content to propose fixes that can change its next actions.