Skip to content

Latest commit

 

History

History
 
 

ChatClientWithTools

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
images
images
 
 
src
src
 
 
README.md
README.md
 
 

README.md

name description page_type languages products urlFragment
.NET MAUI - Chat Client with Tools
Chat UI demonstrating Microsoft.Extensions.AI function calling (tools) with weather (geocode + current), calculator, file listing, system info, and timers.
sample
csharp
xaml
dotnet-maui
chat-client-with-tools

ChatClientWithTools (MAUI + Microsoft.Extensions.AI Tools)

A .NET MAUI sample showing how to enhance using an LLM via Microsoft.Extensions.AI with function calling. This can improve reliability and repeatability of responses, as well as enable the LLM to answer queries it otherwise could not.

chat with tools image

What you'll learn

  • Supplying a set of AIFunction tools to IChatClient via ChatOptions.Tools
  • Implementing strongly described tool schemas (JSON schema) for better LLM argument selection
  • Resolving user-friendly locations to coordinates with a geocode API, then calling a weather endpoint
  • Building rich MAUI UI cards for different tool result types (weather, files, timers, etc.)
  • Safe fallbacks (mock data) when external APIs or permissions are missing

Prerequisites

  • .NET SDK matching this repo branch
  • Azure AI Foundry (or Azure OpenAI) endpoint + API key (model supporting tools/function calling)
  • Optional: OpenWeatherMap API key (for real weather; otherwise mock data)

Environment variables (desktop convenience)

Set these for your user account (PowerShell examples):

[System.Environment]::SetEnvironmentVariable('AZURE_OPENAI_ENDPOINT','https://your-endpoint','User')
[System.Environment]::SetEnvironmentVariable('AZURE_OPENAI_API_KEY','your-api-key','User')
[System.Environment]::SetEnvironmentVariable('AZURE_OPENAI_MODEL','gpt-4o-mini','User')  # optional
[System.Environment]::SetEnvironmentVariable('WEATHER_API_KEY','your-openweather-key','User') # optional

Restart IDE/terminal so the process sees new values.

Tools implemented

Tool Name Purpose Notes
WeatherTool get_weather Current conditions for a location Geocodes first (direct geocoding API) then queries weather by lat/lon; mock fallback w/ note
CalculatorTool calculate Evaluate arithmetic / percentages Sanitizes expression & returns formatted result
FileOperationsTool list_files List files & folders in a common or given path Limits count; resolves shortcuts (Documents, Desktop, Downloads)
SystemInfoTool get_system_info Battery, storage, memory, device info Uses safe simulated values on unsupported platforms
TimerTool set_timer Create a one‑shot timer with title Keeps in-memory timers; prints completion to console

Each tool subclasses AIFunction and overrides: Name, Description, JsonSchema, and InvokeCoreAsync (returning a serializable object for binding).

Key files

  • Services/HostingExtensions.cs – Registers IChatClient (AzureOpenAIClient under the hood) and all tool singletons.
  • Services/Tools/*.cs – Tool implementations (plain AIFunction subclasses; no custom base abstraction to keep things clear).
  • ViewModels/ChatViewModel.cs – Collects tools and invokes _chatClient.GetResponseAsync passing ChatOptions.Tools.
  • Models/ChatMessage.cs – Chat + tool result model with typed helper accessors and UI flags.
  • MainPage.xaml – Chat layout, rich cards per tool result, EmptyView sample prompts, and a Clear toolbar item (resets to EmptyView).

Sample prompts

Try:

  • "What's the weather in Seattle, Washington?"
  • "Calculate 15% tip on $47.50"
  • "List the files in my Documents folder"
  • "Show me current battery level and available storage"
  • "Set a 5-minute timer for my coffee break"
  • "What's 25 * 18 + 150?"
  • "Check the weather in Tokyo and London"

Running (Windows desktop)

Build target net10.0-windows10.0.19041.0 and run. Use a sample prompt or type your own. Use the toolbar "Clear" to reset and revisit the instructional prompt list.

Architecture highlights

Tool invocation flow

  1. User sends natural language input.
  2. ChatViewModel builds ChatOptions with the registered tool objects.
  3. IChatClient (Microsoft.Extensions.AI) decides if any tool(s) should be invoked and executes them, passing JSON arguments that conform to each tool's declared schema.
  4. Tool result objects surface in the assistant response stream and are bound to specialized UI cards.
  5. UI shows structured cards or plain text depending on ChatMessage flags.

Reliability & fallback

  • Weather: If geocode or weather fetch fails (or no API key), returns deterministic mock data with a short annotation so the user knows it's simulated.
  • System info & battery: Returns simulated ranges on unsupported platforms so the UI still illustrates the concept.
  • File listing: Restricts to user directories and caps entries to prevent large payloads.

UI features

  • EmptyView with categorized starter buttons (Weather / Calculator / Files / System Info / Timers).
  • Rich result cards styled with color-coded borders.
  • Toolbar "Clear" to wipe conversation state quickly (teaching scenario friendly).
  • Auto-scroll to the latest message.

Useful references

Notes & disclaimers

  • Requires a model that supports tool (function) calling. Adjust AZURE_OPENAI_MODEL as needed.
  • Weather and other external APIs may rate limit; mock fallbacks keep the UX consistent.
  • Timers are in-memory only (lost on app exit) and currently surface completion via console log.
  • File operations intentionally conservative; expand with care for security.
  • This sample emphasizes pedagogy over exhaustive production hardening.

If you want a lighter starting point without tools, see the SimpleChatClient sample in the same folder tree.