android-memory-evidence

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

• Third-party content exposure detected (high risk: 0.75). The runtime builds the LLM context by scanning the user-supplied QA handoff directory for prior-analysis Markdown/text files (outsider-authored free text), then embeds summaries of that text into analysis_history.previous_analysis_reports / previous_contexts and the top-level evidence.analysis_history in runtime/android_memory_ai/context.py.