my-work
Pass
Audited by Gen Agent Trust Hub on Aug 30, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The script executes the
azCLI tool via a subprocess to obtain an Entra ID access token when a Personal Access Token (PAT) is not provided. This is a standard and documented method for authenticating to Azure services and is restricted to the specificaccount get-access-tokencommand. - [INDIRECT_PROMPT_INJECTION]: The skill processes data from an external source (Azure DevOps work items) which could theoretically contain malicious instructions designed to influence the AI agent.
- Ingestion points: Work item titles, types, and states are retrieved from the Azure DevOps REST API in
scripts/my-work.csand printed to the terminal. - Boundary markers: The data is presented within a text-based table structure, but no explicit 'ignore instructions' delimiters are used for the content of the work items.
- Capability inventory: The skill's primary capabilities are limited to HTTP requests to
dev.azure.comand displaying text to the console. It does not possess file-writing or persistent execution capabilities. - Sanitization: No escaping or sanitization is performed on the strings retrieved from the API before they are displayed.
Audit Metadata