detecting-doc-bloat

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

• Third-party content exposure detected (medium risk: 0.30). In SKILL.md and scripts/plan-chunks.py, the required runtime workflow renders prompts that include the exact contents of documents from the work-repo slice (“doc_lines” / “Chunk {id}: {doc_lines}”) and then the chunk executor writes verdict JSON based on that inline free text.