whatsapp

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

• Third-party content exposure detected (high risk: 0.75). Outsider-authored free text from WhatsApp (e.g., inbound message.text and other event payload fields) is printed as JSONL to stdout in scripts/wa.py’s listen path (_emit → print(json.dumps(_event_payload(...)))), which an agent can then ingest into its LLM context, creating an indirect prompt-injection risk.