retrospective

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

• Third-party content exposure detected (medium risk: 0.65). Multi-session mode reads and parses JSONL session transcripts from ~/.claude/projects/*/*.jsonl, extracting user/assistant message message.content text blocks into the LLM context; those transcripts can include outsider-authored free text (e.g., other participants’ messages) that the operating user did not choose to introduce.