kitaru-hosted-onboarding-tour

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

• Third-party content exposure detected (medium risk: 0.30). The skill instructs the agent to import traces from a checked-in file (traces/langfuse-traces.jsonl) into the Kitaru CLI/runner. These trace files originate from outside the agent's control, representing outsider-authored free text (conversations and customer requests) that the LLM ingests at runtime via session import and inspection.