autoskill

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

• Third-party content exposure detected (high risk: 0.75). At runtime, scripts/fetch_window.py pulls content.text and content.window_name from the local screenpipe /search API (user’s captured OCR/UI text), which is outsider-authored free text originating from other apps/windows and is then inserted into the LLM prompt via scripts/run.py → scripts/synthesize.py (_build_prompt uses cluster example_titles derived from those window titles).