Can I use custom tools with my hosted agent?

Yes, the skill supports adding tools like the built-in Code Interpreter, File Search, and custom MCP (Model Context Protocol) servers.

What are the resource limits for hosted agent containers?

You can allocate between 0.5 to 4 CPU cores and 1Gi to 8Gi of memory per agent instance, with defaults set to 1 CPU and 2Gi of RAM.

What is the minimum SDK version required for this skill?

You must use azure-ai-projects version 2.0.0b3 or later to support hosted agent functionality.

Where should my agent images be stored?

Images must be stored in an Azure Container Registry (ACR), and you must grant the project's managed identity the AcrPull role to access them.

How do I handle authentication for Azure AI Projects?

The skill follows best practices by using DefaultAzureCredential for secure, identity-based access to Azure resources via the azure-identity package.

Azure AI Hosted Agents (Python)

Name: Azure AI Hosted Agents (Python)
Author: microsoft

bymicrosoft

•

1,777

•

Cloud Infrastructure

Builds and deploys container-based hosted agents using the Azure AI Projects SDK for custom code execution in Azure AI Foundry.

The Azure AI Hosted Agents (Python) skill enables developers to create, manage, and deploy specialized containerized agents within the Azure AI Foundry ecosystem. By leveraging the ImageBasedHostedAgentDefinition, users can define custom execution environments with specific CPU and memory allocations, integrate tools like Code Interpreter and MCP servers, and manage agent versioning. This skill is ideal for workflows requiring persistent, sandboxed, or specialized compute environments for AI-driven tasks, providing a standardized path from local development to cloud-hosted agent deployment.

Key Features

01Container-based agent deployment using Azure Container Registry (ACR)

02Native support for Code Interpreter and Model Context Protocol (MCP) tools

03Granular resource allocation for CPU and memory (up to 4 cores/8Gi)

04Comprehensive version management including create, list, and delete operations

05Secure authentication using DefaultAzureCredential and environment variables

061,777 GitHub stars

Use Cases

01Deploying custom Python environments for specialized data processing agents

02Scaling agentic workflows using hosted containers in Azure AI Foundry

03Creating sandboxed execution environments for AI-driven code interpretation

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add microsoft/skills hosted-agents-v2-py

For use in Claude.ai and ChatGPT

Download Skill