project-workspace
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied project names, validation themes, and meeting notes to generate folder structures and markdown documentation (e.g., README.md, validation-plan.md). While this represents a surface for ingesting untrusted data, the agent's capabilities are limited to local file system organization and document generation within the workspace, with no dangerous execution of the processed content.
- [EXTERNAL_DOWNLOADS]: The skill instructions advise fetching product facts and pricing data from official Microsoft Learn and the Azure Retail Prices API. These are well-known, trusted services for retrieving reference information and do not involve executing untrusted code.
- [DATA_EXFILTRATION]: No exfiltration patterns were detected. The skill includes specific guardrails instructing the agent not to store secrets, credentials, or machine-specific absolute paths in the project files.
Audit Metadata