fabriciq-ontology-consumption-cli

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

• Third-party content exposure detected (high risk: 0.75). The required runtime workflow fetches the ontology definition from the Fabric control-plane (getDefinition LRO → decode parts → reconstruct grounding JSON), and that decoded JSON is then placed into the agent’s LLM context; since the ontology content is authored by someone other than the operating user, it is outsider-authored free text/JSON that can carry indirect prompt-injection payloads.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill performs runtime REST calls to the Fabric control plane (notably GET/POST to endpoints such as https://api.fabric.microsoft.com/v1/workspaces/${WS_ID}/items/${ONT_ID}/getDefinition and the operations poll URL) to fetch base64-encoded ontology parts which are decoded into grounding JSON that directly controls routing and query-generation logic used by the agent.