knowledge-compiler

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

• Third-party content exposure detected (high risk: 0.75). Outsider free text can enter the LLM context via the runtime-extracted contents of the user-supplied source file (e.g., PDF/EPUB HTML body text) that extract_source.py writes to cache/jobs/<job-id>/source.txt, which is then read by outline_scout.py/segment_text.py and passed to the agent/subagent for structure inference.