dig

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

• Third-party content exposure detected (medium risk: 0.30). In dig.py, the required workflow reads outsider-authored free text at runtime by loading user-provided session .jsonl files from PROJECT_DIRS and extracting user message text/summary fields (for line in fh: json.loads(line) then t == 'user' text collection and t == 'summary'), which can be attacker-controlled if an outsider can cause/submit poison into those session files.