Fix preserveReasoning flag to control API reasoning inclusion (#9453)

* feat: store reasoning in conversation history for all providers

* refactor: address review feedback

- Move comments inside else block
- Combine reasoning checks into single if block
- Make comments more concise

* refactor: make comments more concise

* Fix preserveReasoning flag to control API reasoning inclusion

Changes:
1. Removed hardcoded <think> tag logic in streaming
   - Previously hardcoded reasoning into assistant message text
   - Now passes reasoning to addToApiConversationHistory as parameter

2. Updated buildCleanConversationHistory to respect preserveReasoning flag
   - When preserveReasoning: true → reasoning block included in API requests
   - When preserveReasoning: false/undefined → reasoning stripped from API
   - Reasoning stored in history for all cases

3. Added temporary debug logs to base-openai-compatible-provider.ts
   - Shows preserveReasoning flag value
   - Logs reasoning blocks in incoming messages
   - Logs <think> tags in converted messages sent to API

* Fix: Use api.getModel() directly instead of cachedStreamingModel

Addresses review comment: cachedStreamingModel is set during streaming but
buildCleanConversationHistory is called before streaming starts. Using the
cached value could cause stale model info when switching models between requests.

Now directly uses this.api.getModel().info.preserveReasoning to ensure we
always check the current model's flag, not a potentially stale cached value.

* Clean up comments in Task.ts

Removed outdated comment regarding model's preserveReasoning flag.

* fix: remove unnecessary reasoningBlock variable in task reasoning logic

Author: daniel-lxs

src/core/task/Task.ts

@@ -2660,21 +2660,14 @@ export class Task extends EventEmitter<TaskEvents> implements TaskLike {
 						})
 					}
 
-					// Check if we should preserve reasoning in the assistant message
-					let finalAssistantMessage = assistantMessage
-					if (reasoningMessage && streamModelInfo.preserveReasoning) {
-						// Prepend reasoning in XML tags to the assistant message so it's included in API history
-						finalAssistantMessage = `<think>${reasoningMessage}</think>\n${assistantMessage}`
-					}
-
 					// Build the assistant message content array
 					const assistantContent: Array<Anthropic.TextBlockParam | Anthropic.ToolUseBlockParam> = []
 
 					// Add text content if present
-					if (finalAssistantMessage) {
+					if (assistantMessage) {
 						assistantContent.push({
 							type: "text" as const,
-							text: finalAssistantMessage,
+							text: assistantMessage,
 						})
 					}
 
@@ -3439,15 +3432,24 @@ export class Task extends EventEmitter<TaskEvents> implements TaskLike {
 
 					continue
 				} else if (hasPlainTextReasoning) {
-					// Strip plain text reasoning, send assistant message only
+					// Check if the model's preserveReasoning flag is set
+					// If true, include the reasoning block in API requests
+					// If false/undefined, strip it out (stored for history only, not sent back to API)
+					const shouldPreserveForApi = this.api.getModel().info.preserveReasoning === true
 					let assistantContent: Anthropic.Messages.MessageParam["content"]
 
-					if (rest.length === 0) {
-						assistantContent = ""
-					} else if (rest.length === 1 && rest[0].type === "text") {
-						assistantContent = (rest[0] as Anthropic.Messages.TextBlockParam).text
+					if (shouldPreserveForApi) {
+						// Include reasoning block in the content sent to API
+						assistantContent = contentArray
 					} else {
-						assistantContent = rest
+						// Strip reasoning out - stored for history only, not sent back to API
+						if (rest.length === 0) {
+							assistantContent = ""
+						} else if (rest.length === 1 && rest[0].type === "text") {
+							assistantContent = (rest[0] as Anthropic.Messages.TextBlockParam).text
+						} else {
+							assistantContent = rest
+						}
 					}
 
 					cleanConversationHistory.push({