write-transcript

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

• Third-party content exposure detected (high risk: 0.70). The required workflow (“Read the entire transcript” and transform it per document-contract) will ingest the conversation transcript text into the agent’s LLM context; if that transcript contains outsider-authored messages (e.g., other participants’ chat content), that free text becomes part of the prompt, creating indirect prompt-injection risk.