dsh-az

Managing Azure resources usually means hopping into a terminal, hand-typing az commands, and then scanning walls of plain-text output for the field you actually need. dsh-az removes that friction by wrapping Azure CLI as native DeepSeek Harness tools, so a dsh agent can query, display, deploy, and diagnose Azure resources directly — collapsing a human-driven command loop into a model-driven, structured one.

The plugin ships seven native tools that cover the full resource lifecycle: az_status checks subscription health, az_resource_query searches resources, az_show inspects details, az_activity_log tracks operation history, and az_group_create plus az_deploy handle provisioning via Bicep or ARM JSON templates. Every query returns clean JSON through --output json, letting the model parse structured data without wrestling with human-formatted text. Write operations — resource-group creation and deployment — always require explicit confirmation before execution, keeping destructive actions safe by design.

If you already rely on dsh agents for development and maintain workloads on Azure, dsh-az compresses the open-terminal, type-command, read-output cycle into a single conversation. As a community-maintained, MIT-licensed project that fits the dsh Microsoft-tooling family alongside SkillOpt, winget, and wsl, it is a drop-in addition for anyone bridging AI-driven workflows with Azure infrastructure.