deploying-to-azure

Pass

Audited by Gen Agent Trust Hub on Mar 9, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill defines a delegation pattern for subagents to analyze deployment logs, creating an indirect prompt injection surface where untrusted log data could influence subagent behavior.\n
  • Ingestion points: The subagent prompt pattern in SKILL.md ingests container logs retrieved via the az containerapp logs show command.\n
  • Boundary markers: The prompt template includes instructions to limit the subagent's response length and focus on specific findings, which provides basic task grounding.\n
  • Capability inventory: The skill utilizes powerful CLI tools including az and azd, as well as PowerShell hooks for environment configuration and resource management.\n
  • Sanitization: No explicit sanitization or filtering of the log content is mentioned before it is processed by the subagent.\n- [COMMAND_EXECUTION]: The skill contains various Azure CLI (az) and Azure Developer CLI (azd) command patterns for infrastructure management, deployment, and troubleshooting. These are standard operations for the skill's intended purpose of Azure environment management.\n- [EXTERNAL_DOWNLOADS]: The Dockerfile examples reference official base images from mcr.microsoft.com and node:22-alpine. These downloads are from well-known and trusted technology providers and are appropriate for building and deploying containerized applications.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 9, 2026, 08:16 PM