ratel-langsmith-analyze

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

• Third-party content exposure detected (high risk: 0.85). Outsider-authored free text is ingested when the skill hydrates LangSmith run trees (Step 4) via read_run(..., load_child_runs=True) / MCP equivalent, because LangSmith stores agent inputs/outputs and transcripts authored by external users and other parties, which the skill then reads and uses as “Evidence” in the LLM context.