deploying-on-azure
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides implementation patterns for Retrieval-Augmented Generation (RAG) which ingests external data into the agent's prompt context.
- Ingestion points: Data is retrieved from Azure AI Search in
examples/sdk/python/azure-openai-rag.pyandreferences/ai-integration.md. - Boundary markers: The implementation lacks strong delimiters or explicit instructions to ignore embedded commands within retrieved content, relying on simple text headers like
[Document 1]. - Capability inventory: The skill facilitates infrastructure deployment via Bicep/Terraform (e.g.,
examples/bicep/container-apps/main.bicep) and data operations via Python SDKs. - Sanitization: No sanitization or filtering of the retrieved search content is performed before interpolation into the chat prompt.
- [EXTERNAL_DOWNLOADS]: The skill references and provides instructions for installing standard, well-known libraries from official registries.
- Evidence: References official Azure and OpenAI packages for Python (
azure-identity,azure-search-documents,openai) and Node.js (@azure/identity,@azure/openai) inreferences/ai-integration.md.
Audit Metadata