plugins/azure/skills/microsoft-foundry

What problem does it solve?

This Skill provides a comprehensive framework for deploying, managing, and troubleshooting Azure AI Foundry resources and agents, simplifying the onboarding and operational processes for AI projects.

Core Features & Use Cases

• Resource Provisioning: Create and configure Azure AI Foundry projects, resources, and networks via Bicep templates and CLI.
• Agent Management: Deploy, update, and troubleshoot both prompt and hosted agents using MCP tools and SDKs.
• Model Deployment: Deploy models with capacity planning, optimal region selection, and SKU configuration, including PTU and Dynamic Quota.
• Evaluation & Monitoring: Auto-create evaluators, run batch evaluations, analyze failures, and set up continuous evaluation workflows.
• Quota & Capacity: Monitor quotas, request increases, and optimize deployment locations to ensure resource availability.
• Networking & Private VNet: Configure private network segmentation for secure deployments.
• Troubleshooting & Telemetry: Use Application Insights and trace analysis for diagnostic insights into agent failures and latency issues.

Quick Start

To set up a new Foundry project, first create the resource with resource/create, then deploy an agent with agent/deploy, and finally run evaluation via observe. Manage quotas with quota. Troubleshoot issues using troubleshoot.