model-office

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

• Third-party content exposure detected (medium risk: 0.30). The required runtime workflow ingests outsider-authored free text directly from the browser into POST /api/task (scripts/server.py) and then passes it to the dispatcher LLM for decomposition and worker prompts (scripts/dispatcher.py::dispatch/decompose, scripts/worker.py::run_subtask), making indirect prompt injection possible.