pipelines-release-notes
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data from Azure DevOps (commit messages, pull request titles, and work item descriptions) which could contain malicious instructions.
- Ingestion points: Azure DevOps REST API responses retrieved in SKILL.md.
- Boundary markers: The instructions do not define clear delimiters or warnings to separate instructions from the data being summarized in the output.
- Capability inventory: Uses curl for network access and shell interaction.
- Sanitization: No specific sanitization or filtering logic is mentioned for the ingested text.
- [DATA_EXFILTRATION]: The skill interacts with the official Azure DevOps API at dev.azure.com to fetch build data. This is a well-known service. Authentication is handled securely via environment variables (AZURE_DEVOPS_PAT, AZURE_DEVOPS_TOKEN), and the skill explicitly forbids printing or saving these tokens.
- [COMMAND_EXECUTION]: The skill executes curl commands using user-provided input such as project names and build IDs. It includes explicit instructions to URL-encode parameters and escape special characters to mitigate shell injection risks.
Audit Metadata