azure-fabric-data-agents

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

• Third-party content exposure detected (high risk: 0.85). Outsider free text can enter the agent’s LLM context via the runtime chat surface (Fabric portal/M365 Copilot/Teams) where the user’s question is provided to the Microsoft-managed Azure OpenAI Assistants API as message content, and that message is not authored by the operating user’s SKILL.md itself.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.70). The skill's SDK path runs "%pip install fabric-data-agent-sdk" inside a notebook, which fetches and executes remote Python package code (e.g., https://pypi.org/project/fabric-data-agent-sdk/) that is then imported and used to create/update agents and their instructions, so it is a runtime external dependency that executes code.