arbor

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

• Third-party content exposure detected (medium risk: 0.30). The required runtime workflow uses scripts/tree.py observe to print/ingest the hypothesis tree’s stored free-text fields (e.g., objective, hypothesis, insight) and those fields can be populated from user/executor-provided arguments via tree.py init/add-node/set-evidence/propagate without any filtering or selecting a specific trusted item first.