microsoft-foundry
Build AI Agents with Microsoft Foundry
This skill helps developers build, deploy, and manage AI agents and models in Microsoft Azure AI Foundry. It provides comprehensive workflows for agent creation, model deployment, capacity planning, RBAC management, and troubleshooting.
Download the skill ZIP
Upload in Claude
Go to Settings → Capabilities → Skills → Upload skill
Toggle on and start using
Test it
Using "microsoft-foundry". Create a new agent with the GPT-4 model
Expected outcome:
I will create a new AI agent using the Microsoft Agent Framework. Let me first check your Azure Foundry project configuration...
Using "microsoft-foundry". Deploy my agent to production
Expected outcome:
I'll help you deploy the agent to Azure. This involves: 1) Building the container image, 2) Pushing to ACR, 3) Creating the agent definition in Foundry.
Using "microsoft-foundry". Where can I deploy a model with 5000 TPM?
Expected outcome:
I'll check capacity across Azure regions for you. Let me query the available capacity...
Security Audit
SafeAll static findings are false positives. The 1370 flagged patterns are documentation code examples in markdown files, not actual code execution. This is an official Microsoft skill for Azure AI Foundry operations. The backtick syntax in markdown code blocks triggered external_commands alerts, path references in docs triggered filesystem alerts, and .env documentation triggered env_access alerts. No malicious intent or actual security vulnerabilities found.
Risk Factors
⚙️ External commands (5)
🔑 Env variables (3)
📁 Filesystem access (3)
🌐 Network access (3)
Detected Patterns
Quality Score
What You Can Build
Build and Deploy Enterprise AI Agents
Create AI agents using the Microsoft Agent Framework SDK with custom tools and multi-agent workflows, then deploy them to Azure Container Apps for production.
Manage Model Deployments at Scale
Discover available model capacity across Azure regions, deploy models with custom configurations, and manage deployment lifecycles efficiently.
Configure Access Control for AI Resources
Set up RBAC permissions, manage role assignments, and configure managed identities for secure access to Foundry resources.
Try These Prompts
Create a new AI agent using Microsoft Agent Framework. The agent should use the gpt-4o model and have a tool that can search Azure documentation.
Deploy my agent to Azure Container Apps. Use ACR for image storage and configure the agent with environment variables for the project endpoint.
Find available capacity for deploying gpt-4o in the East US region. I need at least 10K TPM.
My agent deployment is failing. Check the container logs and tell me what errors are occurring.
Best Practices
- Use azd (Azure Developer CLI) for project context management and environment variables
- Read sub-skill documents before executing workflows - each has specific steps and validation
- Configure environment variables in .env files for local development before deployment
- Use managed identities instead of embedding credentials in agent configurations
Avoid
- Do not skip reading sub-skill documentation before executing workflows
- Do not hardcode Azure credentials in agent source code - use environment variables
- Do not deploy agents without configuring proper RBAC permissions first
Frequently Asked Questions
What is Microsoft Azure AI Foundry?
Do I need an Azure subscription to use this skill?
Can this skill create new Azure resources?
What programming languages are supported?
How do I troubleshoot deployment failures?
Can I use my own model with this skill?
Developer Details
Author
microsoftLicense
MIT
Repository
https://github.com/microsoft/github-copilot-for-azure/tree/main/plugin/skills/microsoft-foundry/Ref
main