building-chat-interfaces

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). The skill prompt instructs developers on how to inject user access tokens into LLM system prompts and tool arguments for MCP tool authentication, which exposes sensitive tokens directly into the LLM context and potential output.

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

• Third-party content exposure detected (medium risk: 0.30). The required workflow involves a chat interface backend that ingests user messages and chat history from client requests, which represent outsider-authored free text from chat users.