mirror of
https://github.com/microsoft/agent-framework.git
synced 2026-06-16 21:04:09 +08:00
westeyandGitHub
e224f06e60
* Update models used in dotnet samples to gpt-5.4-mini * Fix additional missed sample
40 lines
1.7 KiB
Markdown
40 lines
1.7 KiB
Markdown
# What this sample demonstrates
|
|
|
|
This sample demonstrates how to build a hosted agent that uses local C# function tools — a key advantage of code-based hosted agents over prompt agents. The agent acts as a Seattle travel assistant with a `GetAvailableHotels` tool that simulates querying a hotel availability API.
|
|
|
|
Key features:
|
|
- Defining local C# functions as agent tools using `AIFunctionFactory`
|
|
- Using `AIProjectClient` to discover the OpenAI connection from the Microsoft Foundry project
|
|
- Building a `ChatClientAgent` with custom instructions and tools
|
|
- Deploying to the Foundry Hosted Agent service
|
|
|
|
> For common prerequisites and setup instructions, see the [Hosted Agent Samples README](../README.md).
|
|
|
|
## Prerequisites
|
|
|
|
Before running this sample, ensure you have:
|
|
|
|
1. .NET 10 SDK installed
|
|
2. A Microsoft Foundry Project with a chat model deployed (e.g., gpt-5.4-mini)
|
|
3. Azure CLI installed and authenticated (`az login`)
|
|
|
|
## Environment Variables
|
|
|
|
Set the following environment variables:
|
|
|
|
```powershell
|
|
# Replace with your Microsoft Foundry project endpoint
|
|
$env:AZURE_AI_PROJECT_ENDPOINT="https://your-project.services.ai.azure.com/api/projects/your-project-name"
|
|
|
|
# Optional, defaults to gpt-5.4-mini
|
|
$env:MODEL_DEPLOYMENT_NAME="gpt-5.4-mini"
|
|
```
|
|
|
|
## How It Works
|
|
|
|
1. The agent uses `AIProjectClient` to discover the Azure OpenAI connection from the project endpoint
|
|
2. A local C# function `GetAvailableHotels` is registered as a tool using `AIFunctionFactory.Create`
|
|
3. When users ask about hotels, the model invokes the local tool to search simulated hotel data
|
|
4. The tool filters hotels by price and calculates total costs based on the requested dates
|
|
5. Results are returned to the model, which presents them in a conversational format
|