smart-debugging

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

• Third-party content exposure detected (medium risk: 0.30). The skill’s required workflow is specified as ingesting “stack traces,” “error message,” and related troubleshooting artifacts (including user report text in its example) directly from the debugging context rather than from only trusted first-party provider metadata, so outsider-authored free text can be fed into the LLM via the error/report inputs it analyzes.