News
Azure Developer CLI Adds AI Agent Extension for Foundry Deployments
Azure Developer CLI Adds AI Agent Extension for Foundry Deployments
Microsoft introduced a new azd AI agent extension that lets developers build and deploy Microsoft Foundry agents directly from their local development environments.
The update brings agent workflows into the same command-line tooling used for other Azure development tasks, reducing the number of steps required to get an agent from a prototype to a running service.
The extension adds agent-focused project scaffolding to the Azure Developer CLI, using templates that include Infrastructure as Code definitions for the resources an agent requires. Initializing a project with the azd-ai-starter-basic template sets up the directory structure, Bicep files, environment configuration and a declarative azure.yaml file. Developers can then pull in an agent definition from GitHub or a local path, and the CLI updates configuration files and maps parameters to environment variables automatically.
[Click on image for larger view.] Using the Extension (source: Microsoft).
Once a project is initialized, the extension handles the full deployment workflow. The azd up command provisions the Foundry account and project, sets up model deployments and connections, builds container images when needed and publishes the agent. The deployment process, which typically completes in 5--10 minutes for new projects, creates a callable agent endpoint and links to the Foundry playground for testing.
The tooling also applies secure defaults. Managed identity is configured without manual setup, and role assignments and endpoint security are handled through the template's IaC files. This gives developers a ready-made foundation for production agents without configuring authentication or permissions separately.
The announcement highlights several scenarios supported by the new extension, including conversational assistants, data analysis agents connected to services such as Azure SQL Database or Azure Cosmos DB and multi-agent applications. It also supports enterprise patterns like internal blueprints and CI/CD automation using azd provision and azd deploy.
By bringing project scaffolding, model configuration, container builds and deployment into one workflow, the azd AI agent extension gives .NET and VS Code developers a more direct path from local development to running Foundry agents. The extension is available in public preview and is positioned to streamline how Azure developers build agentic applications.
About the Author
David Ramel is an editor and writer at Converge 360.