filter reasoning tokens out of Edit/Apply output and chat utility calls

core/commands/slash/built-in-legacy/commit.ts

@@ -1,5 +1,5 @@
 import { SlashCommand } from "../../../index.js";
-import { renderChatMessage } from "../../../util/messageContent.js";
+import { renderChatMessageWithoutThinking } from "../../../util/messageContent.js";
 
 const CommitMessageCommand: SlashCommand = {
   name: "commit",
@@ -18,7 +18,7 @@ const CommitMessageCommand: SlashCommand = {
       [{ role: "user", content: prompt }],
       abortController.signal,
     )) {
-      yield renderChatMessage(chunk);
+      yield renderChatMessageWithoutThinking(chunk);
     }
   },
 };

core/commands/slash/built-in-legacy/draftIssue.ts

@@ -1,6 +1,6 @@
 import { ChatMessage, SlashCommand } from "../../../index.js";
 import { removeQuotesAndEscapes } from "../../../util/index.js";
-import { renderChatMessage } from "../../../util/messageContent.js";
+import { renderChatMessageWithoutThinking } from "../../../util/messageContent.js";
 
 const PROMPT = (
   input: string,
@@ -49,7 +49,7 @@ const DraftIssueCommand: SlashCommand = {
       abortController.signal,
     )) {
       body += chunk.content;
-      yield renderChatMessage(chunk);
+      yield renderChatMessageWithoutThinking(chunk);
     }
 
     const url = `${params.repositoryUrl}/issues/new?title=${encodeURIComponent(

core/commands/slash/built-in-legacy/onboard.ts

@@ -3,7 +3,7 @@ import ignore from "ignore";
 import type { FileType, IDE, SlashCommand } from "../../..";
 import { getGlobalContinueIgArray } from "../../../indexing/continueignore";
 import { DEFAULT_IGNORE, gitIgArrayFromFile } from "../../../indexing/ignore";
-import { renderChatMessage } from "../../../util/messageContent";
+import { renderChatMessageWithoutThinking } from "../../../util/messageContent";
 import {
   findUriInDirs,
   getUriPathBasename,
@@ -48,7 +48,7 @@ const OnboardSlashCommand: SlashCommand = {
       [{ role: "user", content: prompt }],
       abortController.signal,
     )) {
-      yield renderChatMessage(chunk);
+      yield renderChatMessageWithoutThinking(chunk);
     }
   },
 };

core/commands/slash/built-in-legacy/review.ts

@@ -1,5 +1,5 @@
 import { ChatMessage, SlashCommand } from "../../../index.js";
-import { renderChatMessage } from "../../../util/messageContent.js";
+import { renderChatMessageWithoutThinking } from "../../../util/messageContent.js";
 
 const prompt = `
      Review the following code, focusing on Readability, Maintainability, Code Smells, Speed, and Memory Performance. Provide feedback with these guidelines:
@@ -47,7 +47,7 @@ const ReviewMessageCommand: SlashCommand = {
       [{ role: "user", content: content }],
       abortController.signal,
     )) {
-      yield renderChatMessage(chunk);
+      yield renderChatMessageWithoutThinking(chunk);
     }
   },
 };

core/diff/util.ts

@@ -1,7 +1,7 @@
 import { distance } from "fastest-levenshtein";
 
 import { ChatMessage } from "../index.js";
-import { renderChatMessage } from "../util/messageContent.js";
+import { renderChatMessageWithoutThinking } from "../util/messageContent.js";
 
 export type LineStream = AsyncGenerator<string>;
 
@@ -108,7 +108,9 @@ export async function* streamLines(
   try {
     for await (const update of streamCompletion) {
       const chunk =
-        typeof update === "string" ? update : renderChatMessage(update);
+        typeof update === "string"
+          ? update
+          : renderChatMessageWithoutThinking(update);
       buffer += chunk;
       const lines = buffer.split("\n");
       buffer = lines.pop() ?? "";

core/llm/index.ts

@@ -35,7 +35,10 @@ import { isAbortError } from "../util/isAbortError.js";
 import { isLemonadeInstalled } from "../util/lemonadeHelper.js";
 import { Logger } from "../util/Logger.js";
 import mergeJson from "../util/merge.js";
-import { renderChatMessage } from "../util/messageContent.js";
+import {
+  renderChatMessage,
+  renderChatMessageWithoutThinking,
+} from "../util/messageContent.js";
 import { isOllamaInstalled } from "../util/ollamaHelper.js";
 import { TokensBatchingService } from "../util/TokensBatchingService.js";
 import { withExponentialBackoff } from "../util/withExponentialBackoff.js";
@@ -964,7 +967,7 @@ export abstract class BaseLLM implements ILLM {
   ) {
     let completion = "";
     for await (const message of this.streamChat(messages, signal, options)) {
-      completion += renderChatMessage(message);
+      completion += renderChatMessageWithoutThinking(message);
     }
     return { role: "assistant" as const, content: completion };
   }

core/llm/llms/Anthropic.ts

@@ -27,7 +27,11 @@ import {
   Usage,
 } from "../../index.js";
 import { safeParseToolCallArgs } from "../../tools/parseArgs.js";
-import { renderChatMessage, stripImages } from "../../util/messageContent.js";
+import {
+  renderChatMessage,
+  renderChatMessageWithoutThinking,
+  stripImages,
+} from "../../util/messageContent.js";
 import { extractBase64FromDataUrl } from "../../util/url.js";
 import { DEFAULT_REASONING_TOKENS } from "../constants.js";
 import { BaseLLM } from "../index.js";
@@ -258,7 +262,7 @@ class Anthropic extends BaseLLM {
   ): AsyncGenerator<string> {
     const messages = [{ role: "user" as const, content: prompt }];
     for await (const update of this._streamChat(messages, signal, options)) {
-      yield renderChatMessage(update);
+      yield renderChatMessageWithoutThinking(update);
     }
   }
 

core/llm/llms/Bedrock.ts

@@ -20,7 +20,10 @@ import { fromNodeProviderChain } from "@aws-sdk/credential-providers";
 import type { CompletionOptions } from "../../index.js";
 import { ChatMessage, Chunk, LLMOptions, MessageContent } from "../../index.js";
 import { safeParseToolCallArgs } from "../../tools/parseArgs.js";
-import { renderChatMessage, stripImages } from "../../util/messageContent.js";
+import {
+  renderChatMessageWithoutThinking,
+  stripImages,
+} from "../../util/messageContent.js";
 import { parseDataUrl } from "../../util/url.js";
 import { BaseLLM } from "../index.js";
 import { PROVIDER_TOOL_SUPPORT } from "../toolSupport.js";
@@ -100,7 +103,7 @@ class Bedrock extends BaseLLM {
   ): AsyncGenerator<string> {
     const messages = [{ role: "user" as const, content: prompt }];
     for await (const update of this._streamChat(messages, signal, options)) {
-      yield renderChatMessage(update);
+      yield renderChatMessageWithoutThinking(update);
     }
   }
 

core/llm/llms/Cloudflare.ts

@@ -1,6 +1,6 @@
 import { streamSse } from "@continuedev/fetch";
 import { ChatMessage, CompletionOptions } from "../../index.js";
-import { renderChatMessage } from "../../util/messageContent.js";
+import { renderChatMessageWithoutThinking } from "../../util/messageContent.js";
 import { BaseLLM } from "../index.js";
 
 export default class Cloudflare extends BaseLLM {
@@ -59,7 +59,7 @@ export default class Cloudflare extends BaseLLM {
       signal,
       options,
     )) {
-      yield renderChatMessage(chunk);
+      yield renderChatMessageWithoutThinking(chunk);
     }
   }
 }

core/llm/llms/Cohere.ts

@@ -5,7 +5,10 @@ import {
   CompletionOptions,
   LLMOptions,
 } from "../../index.js";
-import { renderChatMessage, stripImages } from "../../util/messageContent.js";
+import {
+  renderChatMessageWithoutThinking,
+  stripImages,
+} from "../../util/messageContent.js";
 import { BaseLLM } from "../index.js";
 import { DEFAULT_REASONING_TOKENS } from "../constants.js";
 
@@ -148,7 +151,7 @@ class Cohere extends BaseLLM {
   ): AsyncGenerator<string> {
     const messages = [{ role: "user" as const, content: prompt }];
     for await (const update of this._streamChat(messages, signal, options)) {
-      yield renderChatMessage(update);
+      yield renderChatMessageWithoutThinking(update);
     }
   }
 

core/llm/llms/CustomLLM.ts

@@ -1,5 +1,5 @@
 import { ChatMessage, CompletionOptions, CustomLLM } from "../../index.js";
-import { renderChatMessage } from "../../util/messageContent.js";
+import { renderChatMessageWithoutThinking } from "../../util/messageContent.js";
 import { BaseLLM } from "../index.js";
 
 class CustomLLMClass extends BaseLLM {
@@ -76,7 +76,7 @@ class CustomLLMClass extends BaseLLM {
         if (typeof content === "string") {
           yield content;
         } else {
-          yield renderChatMessage(content);
+          yield renderChatMessageWithoutThinking(content);
         }
       }
     } else {

core/llm/llms/Flowise.ts

@@ -1,7 +1,7 @@
 import socketIOClient, { Socket } from "socket.io-client";
 
 import { ChatMessage, CompletionOptions, LLMOptions } from "../../index.js";
-import { renderChatMessage } from "../../util/messageContent.js";
+import { renderChatMessageWithoutThinking } from "../../util/messageContent.js";
 import { BaseLLM } from "../index.js";
 
 interface IFlowiseApiOptions {
@@ -121,7 +121,7 @@ class Flowise extends BaseLLM {
   ): AsyncGenerator<string> {
     const message: ChatMessage = { role: "user", content: prompt };
     for await (const chunk of this._streamChat([message], signal, options)) {
-      yield renderChatMessage(chunk);
+      yield renderChatMessageWithoutThinking(chunk);
     }
   }
 

core/llm/llms/Gemini.ts

@@ -10,7 +10,10 @@ import {
   ToolCallDelta,
 } from "../../index.js";
 import { safeParseToolCallArgs } from "../../tools/parseArgs.js";
-import { renderChatMessage, stripImages } from "../../util/messageContent.js";
+import {
+  renderChatMessageWithoutThinking,
+  stripImages,
+} from "../../util/messageContent.js";
 import { extractBase64FromDataUrl } from "../../util/url.js";
 import { BaseLLM } from "../index.js";
 import { LlmApiRequestType } from "../openaiTypeConverters.js";
@@ -89,7 +92,7 @@ class Gemini extends BaseLLM {
       signal,
       options,
     )) {
-      yield renderChatMessage(message);
+      yield renderChatMessageWithoutThinking(message);
     }
   }
 

core/llm/llms/OpenAI.ts

@@ -16,7 +16,10 @@ import {
   LLMOptions,
   Tool,
 } from "../../index.js";
-import { renderChatMessage } from "../../util/messageContent.js";
+import {
+  renderChatMessage,
+  renderChatMessageWithoutThinking,
+} from "../../util/messageContent.js";
 import { BaseLLM } from "../index.js";
 import {
   fromChatCompletionChunk,
@@ -430,7 +433,7 @@ class OpenAI extends BaseLLM {
       signal,
       options,
     )) {
-      yield renderChatMessage(chunk);
+      yield renderChatMessageWithoutThinking(chunk);
     }
   }
 

core/llm/llms/VertexAI.ts

@@ -2,7 +2,10 @@ import { AuthClient, GoogleAuth, JWT, auth } from "google-auth-library";
 
 import { streamResponse, streamSse } from "@continuedev/fetch";
 import { ChatMessage, CompletionOptions, LLMOptions } from "../../index.js";
-import { renderChatMessage, stripImages } from "../../util/messageContent.js";
+import {
+  renderChatMessageWithoutThinking,
+  stripImages,
+} from "../../util/messageContent.js";
 import { BaseLLM } from "../index.js";
 
 import { LlmApiRequestType } from "../openaiTypeConverters.js";
@@ -489,7 +492,7 @@ class VertexAI extends BaseLLM {
       signal,
       options,
     )) {
-      yield renderChatMessage(message);
+      yield renderChatMessageWithoutThinking(message);
     }
   }
 

core/util/historyUtils.ts

@@ -5,7 +5,7 @@ import path from "path";
 
 import { languageForFilepath } from "../autocomplete/constants/AutocompleteLanguageInfo.js";
 import { ChatMessage, IDE } from "../index.js";
-import { renderChatMessage } from "../util/messageContent.js";
+import { renderChatMessageWithoutThinking } from "../util/messageContent.js";
 import { getContinueGlobalPath } from "../util/paths.js";
 
 // If useful elsewhere, helper funcs should move to core/util/index.ts or similar
@@ -45,7 +45,7 @@ export function toMarkDown(history: ChatMessage[], time?: Date): string {
   let content = `### [Continue](https://continue.dev) session transcript\n Exported: ${time.toLocaleString()}`;
 
   for (const msg of history) {
-    let msgText = renderChatMessage(msg);
+    let msgText = renderChatMessageWithoutThinking(msg);
     if (!msgText) {
       continue; // Skip messages without content
     }

core/util/messageContent.ts

@@ -17,6 +17,11 @@ export function stripImages(messageContent: MessageContent): string {
     .join("\n");
 }
 
+export function renderChatMessageWithoutThinking(message: ChatMessage): string {
+  if (message.role === "thinking") return "";
+  return renderChatMessage(message);
+}
+
 export function renderChatMessage(message: ChatMessage): string {
   switch (message?.role) {
     case "user":