azure-ai-vision-imageanalysis-py
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- Authentication Best Practices: The skill documentation correctly emphasizes the use of
DefaultAzureCredentialand managed identities. This approach follows security best practices by avoiding hardcoded credentials and facilitating automated secret rotation. - External Resource Integration: The skill installs the official
azure-ai-vision-imageanalysispackage and references well-known Microsoft endpoints (e.g.,aka.ms,cognitiveservices.azure.com). These are standard vendor resources required for the skill's operation. - Data Ingestion Surface: By utilizing OCR and image analysis, the skill ingests external data that could theoretically contain instructions intended to influence the agent (indirect prompt injection). This is an inherent property of computer vision tasks and is handled by the underlying platform's safety guardrails.
- Ingestion points: Image data is ingested through
client.analyze_from_urlandclient.analyzemethods defined inSKILL.md. - Boundary markers: No specific boundary markers or "ignore" instructions are present in the provided code snippets.
- Capability inventory: The skill is capable of performing network requests to Azure services and reading local image files.
- Sanitization: No explicit sanitization of extracted text is shown, which is typical for SDK-level examples where the application layer is expected to handle output validation.
Audit Metadata