power-bi-management

Pass

Audited by Gen Agent Trust Hub on Oct 9, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes the Azure CLI (az) using the Python subprocess.run function. The command is constructed as a list of arguments and executed without a shell, which prevents shell injection vulnerabilities. Additionally, the skill validates all Fabric API URLs against an allowlist (api.fabric.microsoft.com) to prevent authentication tokens from being sent to unauthorized external hosts.
  • [DATA_EXFILTRATION]: While the skill transports Power BI report definitions, it employs safety measures to ensure data is only sent to legitimate Microsoft Fabric endpoints. It explicitly checks the hostname of every URL before making a request, mitigating the risk of data being redirected to an attacker-controlled server.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests report definitions and metadata from external Fabric APIs. To prevent malicious data from compromising the host system, the report_definition_io.py script implements rigorous path validation. The _safe_relative function rejects part paths containing NUL bytes, backslashes, absolute paths, or traversal segments (..), and uses pathlib.Path.resolve() to ensure that all extracted files stay within the intended output directory.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 9, 2026, 03:43 PM