azure-openai-2025
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill demonstrates a pattern for analyzing large documents that creates an attack surface for indirect prompt injection.
- Ingestion points: In
SKILL.md, a Python code block reads content fromlarge_document.txtand interpolates it directly into a chat completion prompt. - Boundary markers: The implementation lacks delimiters or instructions directing the model to ignore potentially malicious commands embedded within the document text.
- Capability inventory: The ingested content is processed by the
AzureOpenAIchat completion API, which could be subverted by instructions hidden within the document. - Sanitization: There is no evidence of content validation, escaping, or filtering before the data is passed to the AI model.
Audit Metadata