logs

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

• Third-party content exposure detected (high risk: 0.75). In both quadlet mode (journalctl --user -u charly-<image>.service) and direct mode (<engine> logs <container>), the skill ingests runtime log text produced by the service/container (i.e., not authored by the operating user), which can include arbitrary free-form content that is then fed into the agent’s LLM context.