fp

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

• Third-party content exposure detected (high risk: 0.75). The only runtime text ingestion described in SKILL.md is the agent explicitly fetching a document (scrapling::fetch) and then reading its /content via fp::get; because the fetched document is public/outsider-authored content, it can introduce arbitrary free text into the LLM context through the pipeline steps.