hosted-agents-v2-py
Pass
Audited by Gen Agent Trust Hub on Sep 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the creation of agents configured with
file_searchandmcptools. These tools allow the agent to process data from external files and remote servers which could contain adversarial instructions. - Ingestion points: External data is ingested into the agent context through the
file_searchandmcptool definitions inSKILL.md. - Boundary markers: The code templates do not include explicit delimiters or instructions to the model to ignore potential commands within the ingested data.
- Capability inventory: Agents deployed with these templates have capabilities including code execution via
code_interpreterand remote API interaction viamcp. - Sanitization: No sanitization or validation logic is provided for the content retrieved by the external tools.
- [EXTERNAL_DOWNLOADS]: The installation instructions fetch official Python packages from standard registries.
- Packages:
azure-ai-projectsandazure-identity. - Source: These are official, well-known libraries provided by Microsoft for Azure AI service integration.
Audit Metadata