skills.rest
stephschofield
stephschofieldCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

microsoft-foundry

Deploy, evaluate, and manage Azure AI Foundry agents and model deployments end-to-end.

Updated Jan 23, 2026
Search Tags:#evaluation#rbac#container registry#model deployment#agent deployment#azure cli#azure ai foundry
One-click install
npx skills add https://github.com/stephschofield/beth --skill microsoft-foundry-stephschofield
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: microsoft-foundry
Source: https://github.com/stephschofield/beth/tree/main/.github/skills/microsoft-foundry
Command: npx skills add https://github.com/stephschofield/beth --skill microsoft-foundry-stephschofield

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires azure-ai-projects, azure-identity, and includes references (resource) components.

What problem does it solve? Managing the full lifecycle of Azure AI Foundry agents—containerizing code, pushing images to ACR, creating hosted or prompt agents, running evaluations, and diagnosing failures—requires coordinating many Azure CLI commands, MCP tools, and configuration files. This Skill orchestrates those workflows so agents can be deployed, invoked, evaluated, and troubleshot without manually stitching together each step. ## Core Features & Use Cases - Agent Lifecycle Management: Create hosted agent applications (Microsoft Agent Framework, LangGraph, or custom in Python/C#), deploy prompt or hosted agents, start/stop containers, and clone or delete agents. - Evaluation & Observability: Auto-create evaluators and test datasets after deployment, run batch evaluations, optimize prompts, query traces via Application Insights, and build versioned datasets from production traces. - Infrastructure & Governance: Create Foundry projects and AI Services resources, deploy models with preset/custom/capacity-discovery modes, manage quota and RBAC role assignments, and configure standard or private-network agent setups. - Use Case: A developer asks to deploy a LangGraph agent to Foundry; the Skill downloads a starter sample, generates a Dockerfile, builds and pushes the image to ACR, creates the hosted agent, starts the container, verifies status, and auto-configures evaluators and a test dataset. ## Quick Start Ask the agent to deploy your agent project to Azure AI Foundry, providing your project endpoint and agent name, and it will handle the build, deployment, and verification workflow.

Frequently Asked Questions about microsoft-foundry

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I deploy a hosted agent to Azure AI Foundry?▼

The deploy workflow scans your project type, collects environment variables, generates a Dockerfile, builds the image via ACR cloud build or local Docker, creates the agent with the agent_update MCP tool, and starts the container. It then polls status until Running and auto-creates evaluators and a test dataset.

How do I create a new Azure AI Foundry project?▼

Use the project/create sub-skill, which provisions infrastructure with Azure Developer CLI (azd) using the azd-ai-starter-basic template. It creates the Foundry account, project, Application Insights, managed identity, and RBAC roles, with an optional capability host and Container Registry for hosted agents.

What frameworks are supported for Foundry hosted agents?▼

Hosted agents support Microsoft Agent Framework and LangGraph in Python, plus Microsoft Agent Framework and custom frameworks in C#. LangGraph is Python-only. Starter samples are downloaded from the microsoft-foundry/foundry-samples repository.

Why does my Foundry model deployment fail with QuotaExceeded?▼

QuotaExceeded means your subscription's TPM or PTU limit in that region is exhausted. Check regional usage with az rest against the CognitiveServices usages API, delete unused deployments to free quota immediately, or request an increase through the Azure Portal, which typically takes 3-5 business days.

Can I convert an existing agent project to run on Foundry?▼

Yes, the brownfield workflow wraps your existing agent with a hosting adapter package such as azure-ai-agentserver-agentframework or azure-ai-agentserver-langgraph. You then add an agent.yaml, a Dockerfile targeting linux/amd64 on port 8088, and verify the Responses API endpoint locally.

When should I not use this Skill?▼

Do not use it for Azure Functions, App Service, or general Azure deployments, which belong to dedicated azure-deploy or azure-prepare skills. It is scoped specifically to Microsoft Foundry agents, models, projects, quota, and RBAC.

Related Skills

Explore More Software Engineering Skills→
p-potvinp-potvin
1

microsoft-foundry

Deploy and manage Azure AI Foundry agents across container build, ACR push, hosting, and monitoring.

Community
Advanced
Supporter09Supporter09

langsmith-observability

Trace, evaluate, and monitor LLM application runs with LangSmith.

Community
Advanced
EmeaAppGbbEmeaAppGbb

Azure AI Projects Client Library Skill

Guide Azure AI Projects Python SDK usage for building and deploying AI applications.

Official
Advanced
theslashdojotheslashdojo

functions

Provision and deploy Azure Function Apps via Azure CLI or Core Tools.

Official
Intermediate