[DEV] Implement AzureFoundryService as IAiService provider

[artcava]

Summary

Implement AzureFoundryService as a concrete provider of IAiService, integrated via the existing AiServiceFactory + AiProvider keyed DI architecture. This is a priority item: it must be ready before the Azure MeetUp Torino (see #118), where it will be presented as a real-world Azure AI Foundry integration within XPoster.

---

Context

AiServiceFactory already supports keyed DI resolution by AiProvider enum. OpenAiService is the only active implementation today. AzureFoundryService is the next provider to implement — it does not replace OpenAI but acts as an alternative provider selectable via configuration.

The AiProvider.AzureFoundry enum value is expected to already exist (or must be added) in src/Abstraction/AiProvider.cs.

---

Implementation Plan

1. src/Models/AzureFoundryOptions.cs

Strongly-typed configuration bound from the AzureFoundry configuration section:

public sealed class AzureFoundryOptions
{
    public string Endpoint { get; set; } = string.Empty;
    public string ApiKey { get; set; } = string.Empty;
    public string DeploymentName { get; set; } = string.Empty;
    public string ImageDeploymentName { get; set; } = string.Empty;
    public string ApiVersion { get; set; } = "2024-02-01";
    public double SummaryTemperature { get; set; } = 0.5;
    public int SummaryMaxTokensPerChar { get; set; } = 5;
    public int SummarySafetyMarginChars { get; set; } = 50;
    public string SummarySystemPromptTemplate { get; set; } = "...";
    public string SummaryUserPromptTemplate { get; set; } = "...";
    public string ImagePromptSystemTemplate { get; set; } = "...";
    public string ImagePromptUserTemplate { get; set; } = "...";
    public int ImagePromptMaxTokens { get; set; } = 60;
    public double ImagePromptTemperature { get; set; } = 0.7;
}

Add AzureFoundryOptionsValidator.cs — validate that Endpoint, ApiKey, and DeploymentName are not empty when AiProvider.AzureFoundry is active.

---

2. src/Services/AzureFoundryService.cs

Implement IAiService using the Azure AI Foundry REST API (OpenAI-compatible endpoint):

• GetSummaryAsync — POST to {Endpoint}/openai/deployments/{DeploymentName}/chat/completions?api-version={ApiVersion}
• GetImagePromptAsync — same endpoint, image prompt templates
• GenerateImageAsync — POST to {Endpoint}/openai/deployments/{ImageDeploymentName}/images/generations?api-version={ApiVersion} (DALL-E compatible)

Use IHttpClientFactory. Set api-key header from AzureFoundryOptions.ApiKey.

---

3. Program.cs

builder.Services.AddKeyedTransient<IAiService, AzureFoundryService>(AiProvider.AzureFoundry);
builder.Services.Configure<AzureFoundryOptions>(builder.Configuration.GetSection("AzureFoundry"));
builder.Services.AddSingleton<IValidateOptions<AzureFoundryOptions>, AzureFoundryOptionsValidator>();

---

4. AiServiceFactory.cs

Add AiProvider.AzureFoundry to the supported providers set.

---

5. Configuration files

src/local.settings.json.example:

"AzureFoundry__Endpoint": "",
"AzureFoundry__ApiKey": "",
"AzureFoundry__DeploymentName": "",
"AzureFoundry__ImageDeploymentName": ""

---

6. docs/azure-foundry-setup.md — Setup Guide

Create a step-by-step guide explaining how to provision Azure AI Foundry and retrieve all required parameters to configure AzureFoundryService. The guide must cover:

• How to create an Azure AI Foundry resource in the Azure Portal
• How to create a project and deploy a model (chat completion + image generation)
• How to retrieve the Endpoint URL, API Key, Deployment Name, and API Version
• Where to store these values (App Settings / Key Vault)
• How to switch XPoster to use AzureFoundryService via the AiProvider configuration key
• Troubleshooting: common errors (wrong API version, missing deployment, quota limits)

---

Acceptance Criteria

• AzureFoundryService implements IAiService and compiles
• AzureFoundryOptions bound from AzureFoundry configuration section; no literals hardcoded in the service class
• AiServiceFactory resolves AzureFoundryService when called with AiProvider.AzureFoundry
• AiServiceFactory resolves OpenAiService when called with AiProvider.OpenAi — existing behaviour unchanged
• GetSummaryAsync, GetImagePromptAsync, GenerateImageAsync implemented and functional
• docs/azure-foundry-setup.md written and complete
• All configuration keys documented in local.settings.json.example
• Unit tests pass; CI remains green

---

Tests to Add

tests/Services/AzureFoundryServiceTests.cs:

GetSummaryAsync_WhenTextExceedsLimit_CallsApiAndReturnsTrimmedContent
GetSummaryAsync_WhenApiReturns429_ReturnsEmptyString
GetSummaryAsync_WhenApiReturnsNonSuccess_ReturnsEmptyString
GetImagePromptAsync_WhenApiReturnsValidResponse_ReturnsPrompt
GenerateImageAsync_WhenApiReturnsValidResponse_ReturnsByteArray
GenerateImageAsync_WhenApiReturnsNonSuccess_ReturnsEmptyByteArray

---

Related

• Blocks / feeds into: Azure MeetUp totino #118 (Azure MeetUp Torino — AzureFoundryService must be at minimum scaffolded before the event)
• Architecture reference: AiServiceFactory + keyed DI pattern (see also Add PerplexityService implementation of IAiService using Perplexity Chat Completions API #91 — PerplexityAiService)

[github-actions]

👋 Thanks for opening this issue! A maintainer will review it soon and label it as approved if it can proceed to development.