doctor

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

• Third-party content exposure detected (medium risk: 0.30). Runtime path ingests outsider-authored free text via session transcript JSONL files (~/.claude/projects/<sanitized-cwd>/*.jsonl), which can contain repo web text, tool outputs, and injected instructions, and this skill scans and uses that content at multiple points (e.g., denial/tool-call counting and hook aggregation) without treating it as trusted.