Add structured memory injection and telemetry tracking

- Introduce buildMessagesWithMemory() to prepend memory context as a system
  message before LLM calls
- Record LLM call telemetry (tokens, duration) for follow-up requests in
  both streaming and non-streaming paths
- Add telemetry coverage for tool execution (success/failure, args,
  duration)
- Update tool-executor tests to verify telemetry integration with
  TelemetryManager

src/chat-view.ts

@@ -607,6 +607,18 @@ export class ChatView extends ItemView {
     return filterToolsForMode(allTools, this.currentAgentMode);
   }
 
+  /**
+   * Build messages for the LLM, injecting structured memory context if available.
+   * This is the canonical message builder used for all LLM calls in this view.
+   */
+  private buildMessagesWithMemory(baseMessages: OllamaMessage[]): OllamaMessage[] {
+    const memoryContext = this.structuredMemoryManager?.buildMemoryContext();
+    if (memoryContext) {
+      return [{ role: 'system', content: memoryContext }, ...baseMessages];
+    }
+    return baseMessages;
+  }
+
   buildMessages(userMessageContent: string, tools?: OllamaTool[]): OllamaMessage[] {
     const systemContent = getSystemPromptForMode(this.currentAgentMode);
     const messages: OllamaMessage[] = [];
@@ -730,12 +742,25 @@ export class ChatView extends ItemView {
 
     if (followUpMessages.length > 0) {
       const finalMessages = [...messages, followUp, ...followUpMessages];
+      const followUpStartTime = Date.now();
       const response = await this.ollamaClient.chat(finalMessages, tools);
+      const followUpDurationMs = Date.now() - followUpStartTime;
       const finalResponse = response.content || fullResponse;
       this.updateMessageById(assistantMessageId, {
         content: finalResponse,
         isStreaming: false,
       });
+
+      // Record follow-up LLM call telemetry
+      this.telemetryManager?.recordLlmCall({
+        model: this.settings.model,
+        promptTokens: Math.round(finalMessages.reduce((sum, m) => sum + m.content.length, 0) / 4),
+        completionTokens: Math.round(finalResponse.length / 4),
+        totalTokens: Math.round(
+          (finalMessages.reduce((sum, m) => sum + m.content.length, 0) + finalResponse.length) / 4
+        ),
+        durationMs: followUpDurationMs,
+      });
     }
   }
 
@@ -777,12 +802,25 @@ export class ChatView extends ItemView {
 
     if (followUpMessages.length > 0) {
       const finalMessages = [...messages, followUp, ...followUpMessages];
+      const followUpStartTime = Date.now();
       const response = await this.ollamaClient.chat(finalMessages, tools);
+      const followUpDurationMs = Date.now() - followUpStartTime;
       const finalResponse = response.content || 'Actions applied successfully.';
       this.updateMessageById(assistantMessageId, {
         content: finalResponse,
         isStreaming: false,
       });
+
+      // Record follow-up LLM call telemetry
+      this.telemetryManager?.recordLlmCall({
+        model: this.settings.model,
+        promptTokens: Math.round(finalMessages.reduce((sum, m) => sum + m.content.length, 0) / 4),
+        completionTokens: Math.round(finalResponse.length / 4),
+        totalTokens: Math.round(
+          (finalMessages.reduce((sum, m) => sum + m.content.length, 0) + finalResponse.length) / 4
+        ),
+        durationMs: followUpDurationMs,
+      });
     } else {
       this.updateMessageById(assistantMessageId, {
         content: 'Actions applied successfully.',
@@ -1035,7 +1073,10 @@ export class ChatView extends ItemView {
       const completeMessages =
         this.conversationStateManager.getCompleteMessages(userMessageWithContext);
 
-      const stream = this.ollamaClient.streamChat(completeMessages, tools);
+      // Prepend structured memory as a system message if available
+      const messagesWithMemory = this.buildMessagesWithMemory(completeMessages);
+
+      const stream = this.ollamaClient.streamChat(messagesWithMemory, tools);
 
       let fullResponse = '';
       let toolCalls: OllamaToolCall[] = [];
@@ -1087,7 +1128,7 @@ export class ChatView extends ItemView {
       if (toolCalls.length > 0) {
         await this.processToolCalls(
           toolCalls,
-          completeMessages,
+          messagesWithMemory,
           tools,
           fullResponse,
           assistantMessageId