Edit

Configuring agents in Agent framework with Azure App Configuration

Agents are software systems that autonomously perform tasks using Large Language Models (LLMs) to process user input and execute actions on behalf of users. Agent framework provides tools and structures to build, orchestrate, and manage AI agents. Configuring agents within this framework involves defining their instructions and toolsets through either code-based or declarative methods. Azure App Configuration serves as a centralized store for these agent configurations, keeping them in one place where they can be easily managed and shared across multiple application instances.

Here are some agent settings that can be stored on Azure App Configuration:

  • Instructions
  • Endpoint
  • Agent name
  • Model parameters - temperature, top_p, max_tokens, frequency_penalty, presence_penalty, response_format and stop sequences.
  • Tools

Prerequisites

Example agent settings

  1. Follow the Microsoft Foundry Quickstart to create a project in Foundry with a deployed gpt-5 model. Note down the Foundry project endpoint for later use.

  2. Navigate to your App Configuration store and add the following key-values. Leave Label with its default value. For more information about how to add key-values to a store using the Azure portal or the CLI, go to Create a key-value.

    Key Value Content type
    ChatAgent:ProjectEndpoint Paste the Foundry project endpoint
    ChatAgent:Spec See YAML

    Value for ChatAgent:Spec

    kind: Prompt
name: ChatAgent
description: Agent example with web search
instructions: You are a helpful assistant with access to web search.
model:
    id: gpt-5
    connection:
        kind: remote
tools:
  - kind: web_search
    name: WebSearchTool
    description: Search the web for live information.

  3. Continue to the following instructions to implement the AI agent configuration into your application for the language or platform you're using.

  • Last updated on