chroma

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). The prompt shows examples that pass API keys directly as function arguments (e.g., api_key="your-key"), which instructs embedding secret values into generated code/commands and therefore encourages verbatim handling of credentials.

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

• Third-party content exposure detected (medium risk: 0.65). Outsider free text can enter the agent’s LLM context via the Chroma RAG workflow when the runtime retrieves collection.query(...).documents / vectorstore.similarity_search(...) from a Chroma collection that contains documents added from public web pages, other users’ messages, or downloaded/forwarded files; those retrieved document texts are then typically inserted into the LLM prompt as context.