Added a retrieval context validation agent (#289)

Author: SommerEngineering

README.md

@@ -20,7 +20,7 @@ Things we are currently working on:
   - [ ] Runtime: Integration of the vector database [LanceDB](https://github.com/lancedb/lancedb)
   - [ ] App: Implement the continuous process of vectorizing data
   - [x] ~~App: Define a common retrieval context interface for the integration of RAG processes in chats (PR [#281](https://github.com/MindWorkAI/AI-Studio/pull/281), [#284](https://github.com/MindWorkAI/AI-Studio/pull/284), [#286](https://github.com/MindWorkAI/AI-Studio/pull/286), [#287](https://github.com/MindWorkAI/AI-Studio/pull/287))~~
-  - [x] ~~App: Define a common augmentation interface for the integration of RAG processes in chats (PR [#288](https://github.com/MindWorkAI/AI-Studio/pull/288))~~
+  - [x] ~~App: Define a common augmentation interface for the integration of RAG processes in chats (PR [#288](https://github.com/MindWorkAI/AI-Studio/pull/288), [#289](https://github.com/MindWorkAI/AI-Studio/pull/289))~~
   - [x] ~~App: Integrate data sources in chats (PR [#282](https://github.com/MindWorkAI/AI-Studio/pull/282))~~
 
 

app/MindWork AI Studio/Agents/AgentBase.cs

@@ -11,6 +11,14 @@ namespace AIStudio.Agents;
 
 public abstract class AgentBase(ILogger<AgentBase> logger, SettingsManager settingsManager, DataSourceService dataSourceService, ThreadSafeRandom rng) : IAgent
 {
+    protected static readonly ContentBlock EMPTY_BLOCK = new()
+    {
+        Content = null,
+        ContentType = ContentType.NONE,
+        Role = ChatRole.AGENT,
+        Time = DateTimeOffset.UtcNow,
+    };
+    
     protected static readonly JsonSerializerOptions JSON_SERIALIZER_OPTIONS = new()
     {
         PropertyNamingPolicy = JsonNamingPolicy.SnakeCaseLower,
@@ -24,8 +32,6 @@ public abstract class AgentBase(ILogger<AgentBase> logger, SettingsManager setti
 
     protected ILogger<AgentBase> Logger { get; init; } = logger;
 
-    protected IContent? lastUserPrompt;
-
     /// <summary>
     /// Represents the type or category of this agent.
     /// </summary>
@@ -72,10 +78,10 @@ public abstract class AgentBase(ILogger<AgentBase> logger, SettingsManager setti
         Blocks = [],
     };
 
-    protected DateTimeOffset AddUserRequest(ChatThread thread, string request)
+    protected UserRequest AddUserRequest(ChatThread thread, string request)
     {
         var time = DateTimeOffset.Now;
-        this.lastUserPrompt = new ContentText
+        var lastUserPrompt = new ContentText
         {
             Text = request,
         };
@@ -85,13 +91,17 @@ protected DateTimeOffset AddUserRequest(ChatThread thread, string request)
             Time = time,
             ContentType = ContentType.TEXT,
             Role = ChatRole.USER,
-            Content = this.lastUserPrompt,
+            Content = lastUserPrompt,
         });
-        
-        return time;
+
+        return new()
+        {
+            Time = time,
+            UserPrompt = lastUserPrompt,
+        };
     }
 
-    protected async Task AddAIResponseAsync(ChatThread thread, DateTimeOffset time)
+    protected async Task AddAIResponseAsync(ChatThread thread, IContent lastUserPrompt, DateTimeOffset time)
     {
         if(this.ProviderSettings is null)
             return;
@@ -117,6 +127,6 @@ protected async Task AddAIResponseAsync(ChatThread thread, DateTimeOffset time)
         // Use the selected provider to get the AI response.
         // By awaiting this line, we wait for the entire
         // content to be streamed.
-        await aiText.CreateFromProviderAsync(providerSettings.CreateProvider(this.Logger), providerSettings.Model, this.lastUserPrompt, thread);
+        await aiText.CreateFromProviderAsync(providerSettings.CreateProvider(this.Logger), providerSettings.Model, lastUserPrompt, thread);
     }
 }

app/MindWork AI Studio/Agents/AgentDataSourceSelection.cs

@@ -12,14 +12,6 @@ namespace AIStudio.Agents;
 
 public sealed class AgentDataSourceSelection (ILogger<AgentDataSourceSelection> logger, ILogger<AgentBase> baseLogger, SettingsManager settingsManager, DataSourceService dataSourceService, ThreadSafeRandom rng) : AgentBase(baseLogger, settingsManager, dataSourceService, rng)
 {
-    private static readonly ContentBlock EMPTY_BLOCK = new()
-    {
-        Content = null,
-        ContentType = ContentType.NONE,
-        Role = ChatRole.AGENT,
-        Time = DateTimeOffset.UtcNow,
-    };
-    
     private readonly List<ContentBlock> answers = new();
 
     #region Overrides of AgentBase
@@ -119,8 +111,8 @@ public override async Task<ContentBlock> ProcessInput(ContentBlock input, IDicti
             return EMPTY_BLOCK;
 
         var thread = this.CreateChatThread(this.SystemPrompt(availableDataSources));
-        var time = this.AddUserRequest(thread, text.Text);
-        await this.AddAIResponseAsync(thread, time);
+        var userRequest = this.AddUserRequest(thread, text.Text);
+        await this.AddAIResponseAsync(thread, userRequest.UserPrompt, userRequest.Time);
 
         var answer = thread.Blocks[^1];
 
@@ -306,7 +298,7 @@ public async Task<List<SelectedDataSource>> PerformSelectionAsync(IProvider prov
                 // We know how bad LLM may be in generating JSON without surrounding text.
                 // Thus, we expect the worst and try to extract the JSON list from the text:
                 //
-                var json = this.ExtractJson(selectedDataSourcesJson);
+                var json = ExtractJson(selectedDataSourcesJson);
 
                 try
                 {
@@ -352,7 +344,7 @@ public async Task<List<SelectedDataSource>> PerformSelectionAsync(IProvider prov
     /// </remarks>
     /// <param name="text">The text that may contain the JSON list.</param>
     /// <returns>The extracted JSON list.</returns>
-    private string ExtractJson(string text) => ExtractJson(text.AsSpan()).ToString();
+    private static string ExtractJson(string text) => ExtractJson(text.AsSpan()).ToString();
 
     /// <summary>
     /// Extracts the JSON list from the given text. The text may contain additional