remembering-conversations

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

• Third-party content exposure detected (high risk: 0.85). Yes—outsider-authored free text from archived conversation JSONL files (user/assistant messages from prior sessions) is read at runtime by tool/src/parser.ts and then used to generate summaries/outputs that are fed back into the agent’s LLM context via the search-agent synthesis workflow.