agent-chief-attention-guard

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

• Third-party content exposure detected (high risk: 0.85). SKILL.md describes runtime ingestion of event “body”/“title” text supplied by external sources (e.g., RSS entries from a fetched public feed, Slack/twitter messages, GitHub PRs) via chief.propose_event(...), which the LLM judge then uses as context—an indirect prompt-injection path from outsider-authored free text.