powerbi-report-management

Pass

Audited by Gen Agent Trust Hub on Sep 11, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • Indirect Prompt Injection Surface: The skill processes complex Power BI report definitions (PBIR) retrieved from external Microsoft Fabric APIs or loaded from local project files. Because this data is interpreted by the agent to perform management tasks, it represents a potential surface for indirect prompt injection.
  • Ingestion points: Data enters the agent's context through the getDefinition API call and by reading local .pbip or .pbir files as described in SKILL.md.
  • Boundary markers: The skill does not currently define explicit boundary delimiters or "ignore instructions" tags for the ingested JSON content, though it does recommend using a companion skill for content authoring.
  • Capability inventory: The skill utilizes shell-based tools including az rest, jq, and base64, and performs local file system operations (read/write) to manage report definitions.
  • Sanitization: There is no explicit sanitization or schema validation mentioned for the incoming JSON payloads before they are processed by the agent.
  • Command Execution: The skill relies on local CLI tools (az, jq, base64) to interact with Microsoft Fabric and process data. These operations are essential for the skill's administrative functionality and use standard, well-known tools.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 11, 2026, 01:44 PM
Security Audit — agent-trust-hub — powerbi-report-management