onedrive

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill includes a hard-coded plaintext password ('securePassword123') in the example sharing request, which requires the LLM to emit that secret verbatim in generated code/requests.

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

• Third-party content exposure detected (medium risk: 0.30). In the OneDrive runtime workflow, the agent can ingest file content by downloading from OneDrive via Microsoft Graph endpoints like /users/{userId}/drive/items/{itemId}/content (and similar path-based /content), where the file text is ultimately outsider-authored if the attacker uploads or otherwise places content in the victim’s OneDrive.