rlm

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

• Third-party content exposure detected (high risk: 0.70). The skill’s runtime path rlm_repl.py init <context_path> reads the user-supplied context file into context['content'], and subsequent exec/write_chunks/subagent calls can pass that file’s text (outsider-authored if the file came from public web content, other people’s messages, issues/PRs, etc.) into the LLM context via chunk files and subagent prompts.