rem-sleep

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

• Third-party content exposure detected (high risk: 0.70). The required workflow reads local session transcripts (e.g., ~/.claude/session-transcripts/*.jsonl) and then uses a “Consolidation prompt” to have the agent extract information from those excerpts; since these transcripts are outsider-authored free text from other people’s messages/comments, this creates an indirect prompt-injection path into the LLM context.