temporal-playground

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

• Third-party content exposure detected (high risk: 0.85). The runtime path POST /prompt accepts arbitrary free-text (prompt field) from the browser UI, stores it in store, and later temporal_watch/temporal_get_prompt returns formatPromptResponse(prompt) to the MCP tool—so any outsider-authored message that reaches the UI can be injected into the LLM context.