Background service for chunking and embedding data

app/MindWork AI Studio.sln.DotSettings

@@ -1,4 +1,8 @@
 <wpf:ResourceDictionary xml:space="preserve" xmlns:x="http://schemas.microsoft.com/winfx/2006/xaml" xmlns:s="clr-namespace:System;assembly=mscorlib" xmlns:ss="urn:shemas-jetbrains-com:settings-storage-xaml" xmlns:wpf="http://schemas.microsoft.com/winfx/2006/xaml/presentation">
+	<!-- The local RAG index store is SQLite, where TEXT is stored dynamically. HasMaxLength() has no
+	     effect there, while picking arbitrary limits for paths or chunk texts would turn into real
+	     storage errors as soon as somebody indexes a deeply nested folder or a long document. -->
+	<s:String x:Key="/Default/CodeInspection/Highlighting/InspectionSeverities/=EntityFramework_002EModelValidation_002EUnlimitedStringLength/@EntryIndexedValue">DO_NOT_SHOW</s:String>
 	<s:String x:Key="/Default/CodeStyle/Naming/CSharpNaming/Abbreviations/=AI/@EntryIndexedValue">AI</s:String>
 	<s:String x:Key="/Default/CodeStyle/Naming/CSharpNaming/Abbreviations/=EDI/@EntryIndexedValue">EDI</s:String>
 	<s:String x:Key="/Default/CodeStyle/Naming/CSharpNaming/Abbreviations/=ERI/@EntryIndexedValue">ERI</s:String>

app/MindWork AI Studio/Agents/AgentDataSourceSelection.cs

@@ -140,13 +140,21 @@ public async Task<List<SelectedDataSource>> PerformSelectionAsync(IProvider prov
         //
 
         // We start with the provider currently selected by the user:
-        var agentProvider = this.SettingsManager.GetPreselectedProvider(Tools.Components.AGENT_DATA_SOURCE_SELECTION, provider.Id, true);
+        var requiredDataSecurity = dataSources.AllowedDataSources.GetRequiredSecurityPolicy();
+        var requiredConfidenceLevel = dataSources.AllowedDataSources.GetRequiredConfidenceLevel();
+        var agentProvider = this.SettingsManager.GetPreselectedProvider(Tools.Components.AGENT_DATA_SOURCE_SELECTION, provider.ConfiguredProviderId, true);
         if (agentProvider == Settings.Provider.NONE)
         {
             logger.LogWarning("No provider is selected for the agent. The agent cannot select data sources.");
             return [];
         }
 
+        if (!agentProvider.AllowsDataSourceAccess(this.SettingsManager, requiredDataSecurity, requiredConfidenceLevel))
+        {
+            logger.LogWarning($"The agent for data source selection uses provider '{agentProvider.InstanceName}' with confidence '{agentProvider.GetConfidenceLevel(this.SettingsManager).GetName()}', but the available data sources require data security '{requiredDataSecurity}' and provider confidence '{requiredConfidenceLevel.GetName()}'. The agent cannot select data sources.");
+            return [];
+        }
+
         // Assign the provider settings to the agent:
         logger.LogInformation($"The agent for the data source selection uses the provider '{agentProvider.InstanceName}' ({agentProvider.UsedLLMProvider.ToName()}, confidence={agentProvider.UsedLLMProvider.GetConfidence(this.SettingsManager).Level.GetName()}).");
         this.ProviderSettings = agentProvider;

app/MindWork AI Studio/Agents/AgentRetrievalContextValidation.cs

@@ -3,6 +3,7 @@
 using AIStudio.Chat;
 using AIStudio.Provider;
 using AIStudio.Settings;
+using AIStudio.Settings.DataModel;
 using AIStudio.Tools.RAG;
 using AIStudio.Tools.Services;
 
@@ -129,19 +130,30 @@ public override async Task<ContentBlock> ProcessInput(ContentBlock input, IDicti
     /// you can set the provider once and then call the validation method in parallel.
     /// </remarks>
     /// <param name="provider">The current LLM provider. When the user doesn't preselect an agent provider, the agent uses this provider.</param>
-    public void SetLLMProvider(IProvider provider)
+    /// <param name="requiredDataSecurity">The data security required by the retrieved data.</param>
+    /// <param name="requiredConfidenceLevel">The minimum provider confidence required by the retrieved data.</param>
+    public bool SetLLMProvider(IProvider provider, DataSourceSecurity requiredDataSecurity = DataSourceSecurity.NOT_SPECIFIED, ConfidenceLevel requiredConfidenceLevel = ConfidenceLevel.NONE)
     {
         // We start with the provider currently selected by the user:
-        var agentProvider = this.SettingsManager.GetPreselectedProvider(Tools.Components.AGENT_RETRIEVAL_CONTEXT_VALIDATION, provider.Id, true);
+        var agentProvider = this.SettingsManager.GetPreselectedProvider(Tools.Components.AGENT_RETRIEVAL_CONTEXT_VALIDATION, provider.ConfiguredProviderId, true);
         if (agentProvider == Settings.Provider.NONE)
         {
             logger.LogWarning("No provider is selected for the agent.");
-            return;
+            this.ProviderSettings = Settings.Provider.NONE;
+            return false;
+        }
+
+        if (!agentProvider.AllowsDataSourceAccess(this.SettingsManager, requiredDataSecurity, requiredConfidenceLevel))
+        {
+            logger.LogWarning($"The agent for retrieval context validation uses provider '{agentProvider.InstanceName}' with confidence '{agentProvider.GetConfidenceLevel(this.SettingsManager).GetName()}', but the retrieved data requires data security '{requiredDataSecurity}' and provider confidence '{requiredConfidenceLevel.GetName()}'. The agent cannot validate retrieval contexts.");
+            this.ProviderSettings = Settings.Provider.NONE;
+            return false;
         }
 
         // Assign the provider settings to the agent:
         logger.LogInformation($"The agent for the retrieval context validation uses the provider '{agentProvider.InstanceName}' ({agentProvider.UsedLLMProvider.ToName()}, confidence={agentProvider.UsedLLMProvider.GetConfidence(this.SettingsManager).Level.GetName()}).");
         this.ProviderSettings = agentProvider;
+        return true;
     }
 
     /// <summary>

app/MindWork AI Studio/Assistants/ERI/AssistantERI.razor

@@ -22,7 +22,7 @@
     </MudListItem>
 </MudList>
 
-<PreviewPrototype ApplyInnerScrollingFix="true"/>
+<PreviewBeta ApplyInnerScrollingFix="true"/>
 <div class="mb-6"></div>
 
 <MudText Typo="Typo.h4" Class="mb-3">