changeset

.changeset/bright-grapes-suffer.md

@@ -1,5 +0,0 @@
----
-"claude-dev": patch
----
-
-Fix minor visual issues with auto-approve menu

.changeset/brown-dots-sip.md

@@ -1,5 +0,0 @@
----
-"claude-dev": patch
----
-
-convert condense command to use grpc

.changeset/brown-sloths-search.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": patch
+---
+
+Add the o4-mini model in the isOminiModel

.changeset/calm-snakes-learn.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": patch
+---
+
+Allow option to collect events to send them in a bundle to avoid sending too many events

.changeset/chatty-planets-love.md

@@ -2,4 +2,4 @@
 "claude-dev": minor
 ---
 
-Add Fireworks API Provider
+add open ai cache to ui

.changeset/chilly-oranges-play.md

@@ -2,4 +2,4 @@
 "claude-dev": patch
 ---
 
-add distinct id bootstrap
+disable autocapture

.changeset/clever-balloons-wave.md

@@ -2,4 +2,4 @@
 "claude-dev": patch
 ---
 
-removing free gemini models
+Update deployer team name

.changeset/clever-glasses-tickle.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": patch
+---
+
+Moved rule file conversions

.changeset/cool-bikes-learn.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": minor
+---
+
+suggested model exists again

.changeset/cuddly-clocks-joke.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": patch
+---
+
+wording

.changeset/curly-fireants-work.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": patch
+---
+
+getRelativePaths protobus migration

.changeset/cyan-donuts-sparkle.md

@@ -1,5 +0,0 @@
----
-"claude-dev": patch
----
-
-downloadMcp protobus migration

.changeset/dry-paws-know.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": minor
+---
+
+fixed bug where certain terminal commands would lock you out of a task

.changeset/dull-beds-wash.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": patch
+---
+
+getOllamaModels protobus migration

.changeset/fair-rats-explode.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": minor
+---
+
+Add FeatureFlagProvider service for the Node.js extension side

.changeset/fifty-parents-clean.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": minor
+---
+
+added copy buttons to task header and assistant messages

.changeset/five-eyes-rescue.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": patch
+---
+
+Increasing file size that can be read by cline

.changeset/flat-papayas-doubt.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": patch
+---
+
+Add org to team affiliation check

.changeset/fluffy-pears-cough.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": patch
+---
+
+fix-bedrock-api

.changeset/funny-rivers-check.md

@@ -1,5 +0,0 @@
----
-"claude-dev": patch
----
-
-Added confirmation of a successful cd to cwd before executing commands in the active terminal

.changeset/gentle-paws-shout.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": patch
+---
+
+Add confirmation dialog to Delete All History button

.changeset/happy-radios-enjoy.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": minor
+---
+
+breakpoint just in system prompt for gemini for OR and cline provider

.changeset/heavy-spoons-ring.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": patch
+---
+
+Fix for markdown copy excessive escape characters

.changeset/hungry-rules-relax.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": minor
+---
+
+fetch cache details from generation endpoint

.changeset/lazy-dolls-smell.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": patch
+---
+
+Bump ollama from 0.5.13 to 0.5.15

.changeset/little-needles-play.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": minor
+---
+
+Add npm script for issue creation

.changeset/long-worms-search.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": minor
+---
+
+Migrate more info section to new docs

.changeset/modern-weeks-knock.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": minor
+---
+
+Only initialize posthog in the webview if the user has opted into telemetry

.changeset/moody-brooms-do.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": patch
+---
+
+refreshOpenRouterModels protobus migration

.changeset/nasty-cameras-do.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": patch
+---
+
+Extend ReasoningEffort to non-o3-mini reasoning models for all providers

.changeset/nasty-taxis-trade.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": patch
+---
+
+Re-enable tests in workflow

.changeset/new-pots-walk.md

@@ -1,5 +0,0 @@
----
-"claude-dev": patch
----
-
-This fix adds the --user-data-dir flag when launching Chrome with the remote debugging port, which resolves the 'Chrome was launched but debug port is not responding' error.

.changeset/nice-pears-hope.md

@@ -1,5 +0,0 @@
----
-"claude-dev": patch
----
-
-toggleFavoriteModel protobus migration

.changeset/old-jeans-applaud.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": patch
+---
+
+requestVsCodeLmModels protobus migration

.changeset/purple-planes-speak.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": minor
+---
+
+Allow the user to form their next message while Cline is taking action

.changeset/quick-needles-smell.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": patch
+---
+
+Adding mistral medium-3 model

.changeset/quiet-cups-tie.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": minor
+---
+
+updated gemini caching for OR and cline provider

.changeset/red-carrots-shout.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": patch
+---
+
+searchFiles protobus migration

.changeset/rich-comics-work.md

@@ -1,5 +0,0 @@
----
-"claude-dev": patch
----
-
-adding activation events so cline is activated when vs code opens

.changeset/rude-jeans-impress.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": minor
+---
+
+Add ability to generate commit message via cline

.changeset/selfish-feet-camp.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": minor
+---
+
+updated OR/cline provider to automate gemini models caching

.changeset/selfish-garlics-lay.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": minor
+---
+
+Add Task Favorites and several proto messages related to tasks

.changeset/serious-cameras-pretend.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": patch
+---
+
+getLmStudioModels protobus migration

.changeset/serious-wolves-trade.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": patch
+---
+
+adding quote reply support

.changeset/shiny-falcons-hope.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": minor
+---
+
+Add enterprise section to new docs

.changeset/shiny-snakes-occur.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": patch
+---
+
+Raise Errors when users try to upload images larger than 7500x7500 pixels

.changeset/shy-goats-deliver.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": patch
+---
+
+Make Previous Updates in the Announcement a dropdown

.changeset/silly-donkeys-judge.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": minor
+---
+
+Migrate prompting section to new docs

.changeset/slimy-icons-sin.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": minor
+---
+
+Migrate running models locally section to new docs

.changeset/smart-otters-admire.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": patch
+---
+
+timestamp visualization

.changeset/sour-kiwis-smash.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": patch
+---
+
+refreshOpenAiModels protobus migration

.changeset/stupid-hotels-brush.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": minor
+---
+
+Allow the user to scroll when Cline is editing a file by disabling auto-scroll when the user scrolls up

.changeset/tame-mirrors-destroy.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": minor
+---
+
+Add image links to docs

.changeset/tasty-parrots-applaud.md

@@ -1,5 +0,0 @@
----
-"claude-dev": patch
----
-
-prevent IME composition Enter from auto‑sending edited message

.changeset/thirty-needles-tie.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": minor
+---
+
+Migrate MCP section to new docs

.changeset/tiny-doors-cover.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": minor
+---
+
+Migrate custom model config section to new docs

.changeset/tricky-grapes-accept.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": patch
+---
+
+Fixed an issue where loading never finished when using an application inference profile for the model ID

.changeset/twenty-cobras-cry.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": patch
+---
+
+Add support for Gemini Implicit Caching

.changeset/two-jobs-notice.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": minor
+---
+
+add ui for windsurf and cursor rules

.changeset/warm-gorillas-beam.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": patch
+---
+
+Batch selection and deletion of tasks in history

.changeset/wicked-seas-drive.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": minor
+---
+
+Migrate getting-started section to new docs

.changeset/witty-starfishes-sneeze.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": minor
+---
+
+Task Timeline

.clinerules/cline-overview.md

@@ -209,7 +209,7 @@ class Task {
         switch (chunk.type) {
           case "text":
             // Parse into content blocks
-            this.assistantMessageContent = parseAssistantMessageV2(chunk.text)
+            this.assistantMessageContent = parseAssistantMessage(chunk.text)
             // Present blocks to user
             await this.presentAssistantMessage()
             break

.vscode/launch.json

@@ -16,28 +16,6 @@
 				"IS_DEV": "true",
 				"DEV_WORKSPACE_FOLDER": "${workspaceFolder}"
 			}
-		},
-		{
-			"name": "Run Extension (Fresh Install Mode)",
-			"type": "extensionHost",
-			"request": "launch",
-			"runtimeExecutable": "${execPath}",
-			"args": [
-				"--profile-temp",
-				"--sync",
-				"off",
-				"--disable-extensions",
-				"--extensionDevelopmentPath=${workspaceFolder}",
-				"${workspaceFolder}"
-			],
-			"outFiles": ["${workspaceFolder}/dist/**/*.js"],
-			"preLaunchTask": "clean-sandbox",
-			"internalConsoleOptions": "openOnSessionStart",
-			"postDebugTask": "stop",
-			"env": {
-				"IS_DEV": "true",
-				"DEV_WORKSPACE_FOLDER": "${workspaceFolder}"
-			}
 		}
 	]
 }

.vscode/tasks.json

@@ -185,12 +185,6 @@
 			"label": "stop",
 			"command": "echo ${input:terminate}",
 			"type": "shell"
-		},
-		{
-			"label": "clean-sandbox",
-			"type": "shell",
-			"dependsOn": ["watch"],
-			"command": "rm -rf .vscode-dev"
 		}
 	],
 	"inputs": [

CHANGELOG.md

@@ -1,56 +1,5 @@
 # Changelog
 
-## [3.15.2]
-
-- Added details to auto approve menu and more sensible default controls
-- Add detailed configuration options for LiteLLM provider
-- Add webview telemetry for users who have opted in to telemetry
-- Update Gemini in OpenRouter/Cline providers to use implicit caching
-- Fix freezing issues during rendering of large streaming text
-- Fix grey screen webview crashes by releasing memory after every diff edit
-- Fix breaking out of diff auto-scroll
-- Fix IME composition Enter auto‑sending edited message
-
-## [3.15.1]
-
--   Fix bug where PowerShell commands weren't given enough time before giving up and showing an error
-
-## [3.15.0]
-
--   Add Task Timeline visualization to tasks (Thanks eomcaleb!)
--   Add cache to ui for OpenAi provider
--   Add FeatureFlagProvider service for the Node.js extension side
--   Add copy buttons to task header and assistant messages
--   Add a more simplified home header was added
--   Add ability to favorite a task, allowing it to be kept when clearing all tasks
--   Add npm script for issue creation (Thanks DaveFres!)
--   Add confirmation dialog to Delete All History button
--   Add ability to allow the user to type their next message into the chat while Cline is taking action
--   Add ability to generate commit message via cline (Thanks zapp88!)
--   Add improvements to caching for gemini models on OpenRouter and Cline providers
--   Add improvements to allow scrolling the file being edited.
--   Add ui for windsurf and cursor rules
--   Add mistral medium-3 model
--   Add option to collect events to send them in a bundle to avoid sending too many events
--   Add support to quote a previous message in chat
--   Add support for Gemini Implicit Caching
--   Add support for batch selection and deletion of tasks in history (Thanks danix800!)
--   Update change suggested models
--   Update fetch cache details from generation endpoint
--   Update converted docs to Mintlify
--   Update the isOminiModel to include o4-mini model (Thanks PeterDaveHello!)
--   Update file size that can be read by Cline, allowing larger files
--   Update defaults for bedrock API models (Thanks Watany!)
--   Update to extend ReasoningEffort to non-o3-mini reasoning models for all providers (Thanks PeterDaveHello!)
--   Update to give error when a user tries to upload an image larger than 7500x7500 pixels
--   Update announcement so that previous updates are in a dropdown
--   Update UI for auto approve with favorited settings
--   Fix bug where certain terminal commands would lock you out of a task
--   Fix only initialize posthog in the webview if the user has opted into telemetry
--   Fix bug where autocapture was on for front-end telemetry
--   Fix for markdown copy excessively escaping characters (Thanks weshoke!)
--   Fix an issue where loading never finished when using an application inference profile for the model ID (Thanks WinterYukky!)
-
 ## [3.14.1]
 
 -   Disables autocaptures when initializing feature flags

docs/more-info/telemetry.mdx

@@ -26,9 +26,9 @@ For complete transparency, you can inspect our [telemetry implementation](https:
 
 ### How to Opt Out
 
-Telemetry in Cline is entirely optional:
+Telemetry in Cline is entirely optional and requires your explicit consent:
 
--   When you update or install our VS Code extension, you'll see a message about our anonymous telemetry
+-   When you update or install our VS Code extension, you'll see a simple prompt: "Help Improve Cline" with Allow or Deny options
 -   You can change your preference anytime in settings
 
 Cline also respects VS Code's global telemetry settings. If you've disabled telemetry at the VS Code level, Cline's telemetry will automatically be disabled as well.

package-lock.json

@@ -1,12 +1,12 @@
 {
 	"name": "claude-dev",
-	"version": "3.15.2",
+	"version": "3.14.1",
 	"lockfileVersion": 2,
 	"requires": true,
 	"packages": {
 		"": {
 			"name": "claude-dev",
-			"version": "3.15.2",
+			"version": "3.14.1",
 			"license": "Apache-2.0",
 			"dependencies": {
 				"@anthropic-ai/bedrock-sdk": "^0.12.4",

package.json

@@ -2,7 +2,7 @@
 	"name": "claude-dev",
 	"displayName": "Cline",
 	"description": "Autonomous coding agent right in your IDE, capable of creating/editing files, running commands, using the browser, and more with your permission every step of the way.",
-	"version": "3.15.2",
+	"version": "3.14.1",
 	"icon": "assets/icons/icon.png",
 	"engines": {
 		"vscode": "^1.84.0"
@@ -40,8 +40,6 @@
 		"llama"
 	],
 	"activationEvents": [
-		"onLanguage",
-		"onStartupFinished",
 		"workspaceContains:evals.env"
 	],
 	"main": "./dist/extension.js",
@@ -237,11 +235,45 @@
 		"configuration": {
 			"title": "Cline",
 			"properties": {
+				"cline.vsCodeLmModelSelector": {
+					"type": "object",
+					"properties": {
+						"vendor": {
+							"type": "string",
+							"description": "The vendor of the language model (e.g. copilot)"
+						},
+						"family": {
+							"type": "string",
+							"description": "The family of the language model (e.g. gpt-4)"
+						}
+					},
+					"description": "Settings for VSCode Language Model API"
+				},
 				"cline.enableCheckpoints": {
 					"type": "boolean",
 					"default": true,
 					"description": "Enables extension to save checkpoints of workspace throughout the task. Uses git under the hood which may not work well with large workspaces."
 				},
+				"cline.disableBrowserTool": {
+					"type": "boolean",
+					"default": false,
+					"description": "Disables extension from spawning browser session."
+				},
+				"cline.modelSettings.o3Mini.reasoningEffort": {
+					"type": "string",
+					"enum": [
+						"low",
+						"medium",
+						"high"
+					],
+					"default": "medium",
+					"description": "Controls the reasoning effort when using an OpenAI reasoning model. Higher values may result in more thorough but slower responses."
+				},
+				"cline.chromeExecutablePath": {
+					"type": "string",
+					"default": null,
+					"description": "Path to Chrome executable for browser use functionality. If not set, the extension will attempt to find or download it automatically."
+				},
 				"cline.preferredLanguage": {
 					"type": "string",
 					"enum": [

proto/browser.proto

@@ -40,8 +40,6 @@ message BrowserSettings {
   Viewport viewport = 1;
   optional string remote_browser_host = 2;
   optional bool remote_browser_enabled = 3;
-  optional string chrome_executable_path = 4;
-  optional bool disable_tool_use = 5;
 }
 
 message UpdateBrowserSettingsRequest {
@@ -49,6 +47,4 @@ message UpdateBrowserSettingsRequest {
   Viewport viewport = 2;
   optional string remote_browser_host = 3;
   optional bool remote_browser_enabled = 4;
-  optional string chrome_executable_path = 5;
-  optional bool disable_tool_use = 6;
 }

proto/build-proto.js

@@ -29,7 +29,6 @@ const serviceNameMap = {
 	task: "cline.TaskService",
 	web: "cline.WebService",
 	models: "cline.ModelsService",
-	slash: "cline.SlashService",
 	// Add new services here - no other code changes needed!
 }
 const serviceDirs = Object.keys(serviceNameMap).map((serviceKey) => path.join(ROOT_DIR, "src", "core", "controller", serviceKey))

proto/mcp.proto

@@ -10,7 +10,6 @@ service McpService {
   rpc toggleMcpServer(ToggleMcpServerRequest) returns (McpServers);
   rpc updateMcpTimeout(UpdateMcpTimeoutRequest) returns (McpServers);
   rpc addRemoteMcpServer(AddRemoteMcpServerRequest) returns (McpServers);
-  rpc downloadMcp(StringRequest) returns (Empty);
 }
 
 message ToggleMcpServerRequest {

proto/models.proto

@@ -15,11 +15,9 @@ service ModelsService {
   // Fetches available models from VS Code LM API
   rpc getVsCodeLmModels(EmptyRequest) returns (VsCodeLmModelsArray);
   // Refreshes and returns OpenRouter models
-  rpc refreshOpenRouterModels(EmptyRequest) returns (OpenRouterCompatibleModelInfo);
+  rpc refreshOpenRouterModels(EmptyRequest) returns (OpenRouterModels);
   // Refreshes and returns OpenAI models
   rpc refreshOpenAiModels(OpenAiModelsRequest) returns (StringArray);
-  // Refreshes and returns Requesty models
-  rpc refreshRequestyModels(EmptyRequest) returns (OpenRouterCompatibleModelInfo);
 }
 
 // List of VS Code LM models
@@ -35,7 +33,7 @@ message VsCodeLmModel {
   string id = 4;
 }
 
-// For OpenRouterCompatibleModelInfo structure in OpenRouterModels
+// For ModelInfo structure in OpenRouterModels
 message OpenRouterModelInfo {
   int32 max_tokens = 1;
   int32 context_window = 2;
@@ -48,8 +46,8 @@ message OpenRouterModelInfo {
   string description = 9;
 }
 
-// Shared response message for model information
-message OpenRouterCompatibleModelInfo {
+// Response message for OpenRouter models
+message OpenRouterModels {
   map<string, OpenRouterModelInfo> models = 1;
 }
 
@@ -59,3 +57,4 @@ message OpenAiModelsRequest {
   string baseUrl = 2;
   string apiKey = 3;
 }
+

proto/slash.proto

@@ -1,14 +0,0 @@
-syntax = "proto3";
-
-package cline;
-option java_package = "bot.cline.proto";
-option java_multiple_files = true;
-
-import "common.proto";
-
-// SlashService provides methods for managing slash
-service SlashService {
-  // Sends button click message
-  rpc reportBug(StringRequest) returns (Empty);
-  rpc condense(StringRequest) returns (Empty);
-}

proto/state.proto

@@ -6,7 +6,6 @@ import "common.proto";
 service StateService {
   rpc getLatestState(EmptyRequest) returns (State);
   rpc subscribeToState(EmptyRequest) returns (stream State);
-  rpc toggleFavoriteModel(StringRequest) returns (Empty);
 }
 
 message State {

src/api/index.ts

@@ -19,7 +19,6 @@ import { DoubaoHandler } from "./providers/doubao"
 import { VsCodeLmHandler } from "./providers/vscode-lm"
 import { ClineHandler } from "./providers/cline"
 import { LiteLlmHandler } from "./providers/litellm"
-import { FireworksHandler } from "./providers/fireworks"
 import { AskSageHandler } from "./providers/asksage"
 import { XAIHandler } from "./providers/xai"
 import { SambanovaHandler } from "./providers/sambanova"
@@ -59,8 +58,6 @@ export function buildApiHandler(configuration: ApiConfiguration): ApiHandler {
 			return new DeepSeekHandler(options)
 		case "requesty":
 			return new RequestyHandler(options)
-		case "fireworks":
-			return new FireworksHandler(options)
 		case "together":
 			return new TogetherHandler(options)
 		case "qwen":

src/api/providers/fireworks.ts

@@ -1,94 +0,0 @@
-import { Anthropic } from "@anthropic-ai/sdk"
-import OpenAI from "openai"
-import { withRetry } from "../retry"
-import { ApiHandler } from ".."
-import {
-	ApiHandlerOptions,
-	DeepSeekModelId,
-	ModelInfo,
-	deepSeekDefaultModelId,
-	deepSeekModels,
-	openAiModelInfoSaneDefaults,
-} from "../../shared/api"
-import { convertToOpenAiMessages } from "../transform/openai-format"
-import { ApiStream } from "../transform/stream"
-
-export class FireworksHandler implements ApiHandler {
-	private options: ApiHandlerOptions
-	private client: OpenAI
-
-	constructor(options: ApiHandlerOptions) {
-		this.options = options
-		this.client = new OpenAI({
-			baseURL: "https://api.fireworks.ai/inference/v1",
-			apiKey: this.options.fireworksApiKey,
-		})
-	}
-
-	@withRetry()
-	async *createMessage(systemPrompt: string, messages: Anthropic.Messages.MessageParam[]): ApiStream {
-		const modelId = this.options.fireworksModelId ?? ""
-
-		const openAiMessages: OpenAI.Chat.ChatCompletionMessageParam[] = [
-			{ role: "system", content: systemPrompt },
-			...convertToOpenAiMessages(messages),
-		]
-
-		const stream = await this.client.chat.completions.create({
-			model: modelId,
-			...(this.options.fireworksModelMaxCompletionTokens
-				? { max_completion_tokens: this.options.fireworksModelMaxCompletionTokens }
-				: {}),
-			...(this.options.fireworksModelMaxTokens ? { max_tokens: this.options.fireworksModelMaxTokens } : {}),
-			messages: openAiMessages,
-			stream: true,
-			stream_options: { include_usage: true },
-			temperature: 0,
-		})
-
-		let reasoning: string | null = null
-		for await (const chunk of stream) {
-			const delta = chunk.choices[0]?.delta
-			if (reasoning || delta?.content?.includes("<think>")) {
-				reasoning = (reasoning || "") + (delta.content ?? "")
-			}
-
-			if (delta?.content && !reasoning) {
-				yield {
-					type: "text",
-					text: delta.content,
-				}
-			}
-
-			if (reasoning || ("reasoning_content" in delta && delta.reasoning_content)) {
-				yield {
-					type: "reasoning",
-					reasoning: delta.content || ((delta as any).reasoning_content as string | undefined) || "",
-				}
-				if (reasoning?.includes("</think>")) {
-					// Reset so the next chunk is regular content
-					reasoning = null
-				}
-			}
-
-			if (chunk.usage) {
-				yield {
-					type: "usage",
-					inputTokens: chunk.usage.prompt_tokens || 0, // (deepseek reports total input AND cache reads/writes, see context caching: https://api-docs.deepseek.com/guides/kv_cache) where the input tokens is the sum of the cache hits/misses, while anthropic reports them as separate tokens. This is important to know for 1) context management truncation algorithm, and 2) cost calculation (NOTE: we report both input and cache stats but for now set input price to 0 since all the cost calculation will be done using cache hits/misses)
-					outputTokens: chunk.usage.completion_tokens || 0,
-					// @ts-ignore-next-line
-					cacheReadTokens: chunk.usage.prompt_cache_hit_tokens || 0,
-					// @ts-ignore-next-line
-					cacheWriteTokens: chunk.usage.prompt_cache_miss_tokens || 0,
-				}
-			}
-		}
-	}
-
-	getModel(): { id: string; info: ModelInfo } {
-		return {
-			id: this.options.fireworksModelId ?? "",
-			info: openAiModelInfoSaneDefaults,
-		}
-	}
-}

src/api/providers/litellm.ts

@@ -65,7 +65,7 @@ export class LiteLlmHandler implements ApiHandler {
 		const reasoningOn = budgetTokens !== 0 ? true : false
 		const thinkingConfig = reasoningOn ? { type: "enabled", budget_tokens: budgetTokens } : undefined
 
-		let temperature: number | undefined = this.options.liteLlmModelInfo?.temperature ?? 0
+		let temperature: number | undefined = 0
 
 		if (isOminiModel && reasoningOn) {
 			temperature = undefined // Thinking mode doesn't support temperature
@@ -169,7 +169,7 @@ export class LiteLlmHandler implements ApiHandler {
 	getModel() {
 		return {
 			id: this.options.liteLlmModelId || liteLlmDefaultModelId,
-			info: this.options.liteLlmModelInfo || liteLlmModelInfoSaneDefaults,
+			info: liteLlmModelInfoSaneDefaults,
 		}
 	}
 }

src/api/transform/openrouter-stream.ts

@@ -75,6 +75,74 @@ export async function createOpenRouterStream(
 			break
 	}
 
+	// handles gemini caching logic
+	if (model.id.startsWith("google/") && model.info.supportsPromptCache) {
+		// gemini only uses the last breakpoint for caching, so the others will be ignored
+		openAiMessages[0] = {
+			role: "system",
+			content: [
+				{
+					type: "text",
+					text: systemPrompt,
+					// @ts-ignore-next-line
+					cache_control: { type: "ephemeral" },
+				},
+			],
+		}
+
+		// for safety, but this should always be the case
+		if (openAiMessages.length >= 2) {
+			const msg = openAiMessages[1]
+
+			if (msg) {
+				if (typeof msg.content === "string") {
+					msg.content = [{ type: "text", text: msg.content }]
+				}
+				if (Array.isArray(msg.content)) {
+					// NOTE: this is fine since env details will always be added at the end. but if it weren't there, and the user added a image_url type message, it would pop a text part before it and then move it after to the end.
+					let lastTextPart = msg.content.filter((part) => part.type === "text").pop()
+
+					if (!lastTextPart) {
+						lastTextPart = { type: "text", text: "..." }
+						msg.content.push(lastTextPart)
+					}
+					// @ts-ignore-next-line
+					lastTextPart["cache_control"] = { type: "ephemeral" }
+				}
+			}
+		}
+
+		// it doesn't make sense to alter breakpoints at all with the gemini cache implementation at this time
+		/*const GEMINI_CACHE_USER_MESSAGE_INTERVAL = 4 // add new breakpoint every 4 turns
+		const userMessages = openAiMessages.filter((msg) => msg.role === "user")
+
+		const userMessageCount = userMessages.length
+		const targetUserMessageNumber =
+			Math.floor(userMessageCount / GEMINI_CACHE_USER_MESSAGE_INTERVAL) * GEMINI_CACHE_USER_MESSAGE_INTERVAL
+
+		if (targetUserMessageNumber > 0) {
+			// otherwise dont need to add a breakpoint
+			const msg = userMessages[targetUserMessageNumber - 1]
+
+			if (msg) {
+				if (typeof msg.content === "string") {
+					msg.content = [{ type: "text", text: msg.content }]
+				}
+				if (Array.isArray(msg.content)) {
+					// NOTE: this is fine since env details will always be added at the end. but if it weren't there, and the user added a image_url type message, it would pop a text part before it and then move it after to the end.
+					let lastTextPart = msg.content.filter((part) => part.type === "text").pop()
+
+					if (!lastTextPart) {
+						lastTextPart = { type: "text", text: "..." }
+						msg.content.push(lastTextPart)
+					}
+					// @ts-ignore-next-line
+					lastTextPart["cache_control"] = { type: "ephemeral" }
+				}
+			}
+		}*/
+	}
+
 	// Not sure how openrouter defaults max tokens when no value is provided, but the anthropic api requires this value and since they offer both 4096 and 8192 variants, we should ensure 8192.
 	// (models usually default to max tokens allowed)
 	let maxTokens: number | undefined

src/core/assistant-message/index.ts

@@ -1,6 +1,6 @@
 export type AssistantMessageContent = TextContent | ToolUse
 
-export { parseAssistantMessageV1, parseAssistantMessageV2 } from "./parse-assistant-message"
+export { parseAssistantMessage } from "./parse-assistant-message"
 
 export interface TextContent {
 	type: "text"

src/core/assistant-message/parse-assistant-message.ts

@@ -1,24 +1,6 @@
-import { AssistantMessageContent, TextContent, ToolUse, ToolParamName, toolParamNames, toolUseNames, ToolUseName } from "." // Assuming types are defined in index.ts or a similar file
+import { AssistantMessageContent, TextContent, ToolUse, ToolParamName, toolParamNames, toolUseNames, ToolUseName } from "."
 
-/**
- * @description **Version 1**
- * Parses an assistant message string potentially containing mixed text and tool usage blocks
- * marked with XML-like tags into an array of structured content objects.
- *
- * This version iterates through the message character by character, building an accumulator string.
- * It maintains state to track whether it's currently parsing text, a tool use block, or a specific tool parameter.
- * It detects the start and end of tool uses and parameters by checking if the accumulator ends with
- * the corresponding opening or closing tags.
- * Special handling is included for `write_to_file` and `new_rule` tool uses to correctly parse
- * the `content` parameter, which might contain the closing tag itself, by looking for the *last*
- * occurrence of the closing tag.
- * If the input string ends mid-tag or mid-content, the last block (text or tool use) is marked as partial.
- *
- * @param assistantMessage The raw string output from the assistant.
- * @returns An array of `AssistantMessageContent` objects, which can be `TextContent` or `ToolUse`.
- *          Blocks that were not fully closed by the end of the input string will have their `partial` flag set to `true`.
- */
-export function parseAssistantMessageV1(assistantMessage: string): AssistantMessageContent[] {
+export function parseAssistantMessage(assistantMessage: string) {
 	const contentBlocks: AssistantMessageContent[] = []
 	let currentTextContent: TextContent | undefined = undefined
 	let currentTextContentStartIndex = 0
@@ -32,56 +14,46 @@ export function parseAssistantMessageV1(assistantMessage: string): AssistantMess
 		const char = assistantMessage[i]
 		accumulator += char
 
-		// --- State: Parsing a Tool Parameter ---
 		// there should not be a param without a tool use
 		if (currentToolUse && currentParamName) {
 			const currentParamValue = accumulator.slice(currentParamValueStartIndex)
 			const paramClosingTag = `</${currentParamName}>`
 			if (currentParamValue.endsWith(paramClosingTag)) {
-				// End of param value found
+				// end of param value
 				currentToolUse.params[currentParamName] = currentParamValue.slice(0, -paramClosingTag.length).trim()
-				currentParamName = undefined // Go back to parsing tool content or looking for next param
-				continue // Move to next character
+				currentParamName = undefined
+				continue
 			} else {
-				// Partial param value is accumulating
-				continue // Move to next character
+				// partial param value is accumulating
+				continue
 			}
 		}
 
-		// --- State: Parsing a Tool Use (but not a specific parameter) ---
 		// no currentParamName
+
 		if (currentToolUse) {
 			const currentToolValue = accumulator.slice(currentToolUseStartIndex)
 			const toolUseClosingTag = `</${currentToolUse.name}>`
-
 			if (currentToolValue.endsWith(toolUseClosingTag)) {
-				// End of a tool use found
+				// end of a tool use
 				currentToolUse.partial = false
 				contentBlocks.push(currentToolUse)
-				currentToolUse = undefined // Go back to parsing text or looking for next tool
-				// Reset text start index in case text follows immediately
-				currentTextContentStartIndex = i + 1
-				continue // Move to next character
+				currentToolUse = undefined
+				continue
 			} else {
-				// Check if starting a new parameter within the current tool use
 				const possibleParamOpeningTags = toolParamNames.map((name) => `<${name}>`)
-				let foundParamStart = false
 				for (const paramOpeningTag of possibleParamOpeningTags) {
 					if (accumulator.endsWith(paramOpeningTag)) {
-						// Start of a new parameter found
+						// start of a new parameter
 						currentParamName = paramOpeningTag.slice(1, -1) as ToolParamName
 						currentParamValueStartIndex = accumulator.length
-						foundParamStart = true
 						break
 					}
 				}
-				if (foundParamStart) {
-					continue // Move to next character
-				}
 
-				// Special case for write_to_file/new_rule content param allowing nested tags
-				// Check if a </content> tag appears, potentially indicating the end of the content param
-				// even if the main tool closing tag hasn't been seen yet.
+				// there's no current param, and not starting a new param
+
+				// special case for write_to_file where file contents could contain the closing tag, in which case the param would have closed and we end up with the rest of the file contents here. To work around this, we get the string between the starting content tag and the LAST content tag.
 				const contentParamName: ToolParamName = "content"
 				if (
 					(currentToolUse.name === "write_to_file" || currentToolUse.name === "new_rule") &&
@@ -91,384 +63,73 @@ export function parseAssistantMessageV1(assistantMessage: string): AssistantMess
 					const contentStartTag = `<${contentParamName}>`
 					const contentEndTag = `</${contentParamName}>`
 					const contentStartIndex = toolContent.indexOf(contentStartTag) + contentStartTag.length
-					// Use lastIndexOf to handle cases where </content> might appear within the content itself
 					const contentEndIndex = toolContent.lastIndexOf(contentEndTag)
-
-					// Ensure we found valid start/end tags and end is after start
-					if (
-						contentStartIndex !== -1 &&
-						contentEndIndex !== -1 &&
-						contentEndIndex > contentStartIndex - contentStartTag.length // Ensure end tag is after start tag begins
-					) {
-						// Check if this content param was already being parsed. If so, update it.
-						// If not, and we just found the closing tag, assign it.
-						// This handles cases where the </content> detection might fire before
-						// the <content> tag detection logic, or if the content is very short.
-						if (currentParamName === contentParamName) {
-							// Already parsing content, now we found the end tag
-							currentToolUse.params[contentParamName] = toolContent.slice(contentStartIndex, contentEndIndex).trim()
-							currentParamName = undefined // Finished with this param
-						} else if (currentParamName === undefined) {
-							// Not parsing a param, but found </content>. Assume it closes the content block.
-							currentToolUse.params[contentParamName] = toolContent.slice(contentStartIndex, contentEndIndex).trim()
-							// We stay in the "parsing tool use" state, looking for more params or the tool end tag.
-						}
+					if (contentStartIndex !== -1 && contentEndIndex !== -1 && contentEndIndex > contentStartIndex) {
+						currentToolUse.params[contentParamName] = toolContent.slice(contentStartIndex, contentEndIndex).trim()
 					}
 				}
 
-				// If none of the above, partial tool value is accumulating
-				continue // Move to next character
+				// partial tool value is accumulating
+				continue
 			}
 		}
 
-		// --- State: Parsing Text (or looking for start of a tool use) ---
 		// no currentToolUse
+
 		let didStartToolUse = false
 		const possibleToolUseOpeningTags = toolUseNames.map((name) => `<${name}>`)
 		for (const toolUseOpeningTag of possibleToolUseOpeningTags) {
 			if (accumulator.endsWith(toolUseOpeningTag)) {
-				// Start of a new tool use found
-				const toolName = toolUseOpeningTag.slice(1, -1) as ToolUseName
+				// start of a new tool use
 				currentToolUse = {
 					type: "tool_use",
-					name: toolName,
+					name: toolUseOpeningTag.slice(1, -1) as ToolUseName,
 					params: {},
 					partial: true,
 				}
 				currentToolUseStartIndex = accumulator.length
-
-				// This also indicates the end of the current text content block (if any)
+				// this also indicates the end of the current text content
 				if (currentTextContent) {
 					currentTextContent.partial = false
-					// Extract text content, removing the part that formed the tool opening tag
-					const textEndIndex = accumulator.length - toolUseOpeningTag.length
-					currentTextContent.content = accumulator.slice(currentTextContentStartIndex, textEndIndex).trim()
-					// Only add if there's actual content
-					if (currentTextContent.content.length > 0) {
-						contentBlocks.push(currentTextContent)
-					}
+					// remove the partially accumulated tool use tag from the end of text (<tool)
+					currentTextContent.content = currentTextContent.content
+						.slice(0, -toolUseOpeningTag.slice(0, -1).length)
+						.trim()
+					contentBlocks.push(currentTextContent)
 					currentTextContent = undefined
-				} else {
-					// Check if there was text before this tool use started
-					const textEndIndex = accumulator.length - toolUseOpeningTag.length
-					const potentialText = accumulator.slice(currentTextContentStartIndex, textEndIndex).trim()
-					if (potentialText.length > 0) {
-						contentBlocks.push({
-							type: "text",
-							content: potentialText,
-							partial: false, // Ended because tool use started
-						})
-					}
 				}
 
 				didStartToolUse = true
-				break // Found tool start, stop checking for others
+				break
 			}
 		}
 
 		if (!didStartToolUse) {
-			// No tool use started, so it must be text content accumulating
-			// (or continuing after a closed tool use)
+			// no tool use, so it must be text either at the beginning or between tools
 			if (currentTextContent === undefined) {
-				// Start of a new text block
-				currentTextContentStartIndex = i - (accumulator.length - currentTextContentStartIndex - 1) // Adjust start index based on how much we've accumulated since the last block ended or the beginning
-				// If accumulator starts from 0, start index is i
-				if (contentBlocks.length === 0 && currentToolUse === undefined) {
-					currentTextContentStartIndex = accumulator.length - 1 // i
-				} else {
-					// Re-calculate based on the actual start of the current text segment
-					// Find the end of the last block
-					let lastBlockEndIndex = 0
-					if (contentBlocks.length > 0) {
-						const lastBlock = contentBlocks[contentBlocks.length - 1]
-						// Approximation: find where the accumulator matches the end of the message string representation of the last block. This is complex.
-						// Simpler: Assume text starts right after the last block ended implicitly at index i.
-						lastBlockEndIndex = i // Where the loop *was* when the last block finished processing
-						// Need a more robust way to track the end index of the *raw string* corresponding to the last block.
-						// Let's stick to the accumulator slice approach for simplicity in this version.
-						// The start index should be where the current *unmatched* text began.
-						let lastProcessedIndex = -1
-						if (contentBlocks.length > 0) {
-							// This requires knowing the raw string length of the previous block, which V1 doesn't explicitly track easily.
-							// We'll approximate based on the current accumulator and start index logic.
-							// The issue arises if a tool tag was just closed. accumulator contains everything up to i.
-							// lastBlockEndIndex should point to the character *after* the closing tag of the last block.
-						}
-						// Reset start index to the beginning of the *current* potential text block
-						currentTextContentStartIndex = accumulator.length - 1 // Start accumulating from the current character `i`
-					}
-
-					// If we just closed a tool, text starts *after* its closing tag
-					// The logic needs refinement here for accurate start index after a tool closure.
-					// Let's assume for now the start index logic inside the loop handles it via slicing.
-				}
-
-				currentTextContent = {
-					type: "text",
-					content: "", // Content will be filled by slicing accumulator
-					partial: true,
-				}
+				currentTextContentStartIndex = i
+			}
+			currentTextContent = {
+				type: "text",
+				content: accumulator.slice(currentTextContentStartIndex).trim(),
+				partial: true,
 			}
-			// Update text content based on the accumulator from its start index
-			currentTextContent.content = accumulator.slice(currentTextContentStartIndex).trimStart() // Trim start to avoid leading space if text follows tool
 		}
-	} // End of loop
+	}
 
-	// --- Finalization after loop ---
-
-	// If a tool use was open at the end
 	if (currentToolUse) {
-		// If a parameter was open within that tool use
+		// stream did not complete tool call, add it as partial
 		if (currentParamName) {
-			// The remaining accumulator content belongs to this partial parameter
+			// tool call has a parameter that was not completed
 			currentToolUse.params[currentParamName] = accumulator.slice(currentParamValueStartIndex).trim()
 		}
-		// Add the potentially partial tool use block
 		contentBlocks.push(currentToolUse)
 	}
-	// If text content was being accumulated at the end
-	// Note: Only one of currentToolUse or currentTextContent can be defined here,
-	// as starting a tool use finalizes the preceding text block.
-	else if (currentTextContent) {
-		// Update content one last time
-		currentTextContent.content = accumulator.slice(currentTextContentStartIndex).trim()
-		// Add the potentially partial text block only if it contains content
-		if (currentTextContent.content.length > 0) {
-			contentBlocks.push(currentTextContent)
-		}
-	}
 
-	return contentBlocks
-}
-
-/**
- * @description **Version 2**
- * Parses an assistant message string potentially containing mixed text and tool usage blocks
- * marked with XML-like tags into an array of structured content objects.
- *
- * This version aims for efficiency by avoiding the character-by-character accumulator of V1.
- * It iterates through the string using an index `i`. At each position, it checks if the substring
- * *ending* at `i` matches any known opening or closing tags for tools or parameters using `startsWith`
- * with an offset.
- * It uses pre-computed Maps (`toolUseOpenTags`, `toolParamOpenTags`) for quick tag lookups.
- * State is managed using indices (`currentTextContentStart`, `currentToolUseStart`, `currentParamValueStart`)
- * pointing to the start of the current block within the original `assistantMessage` string.
- * Slicing is used to extract content only when a block (text, parameter, or tool use) is completed.
- * Special handling for `write_to_file` and `new_rule` content parameters is included, using `indexOf`
- * and `lastIndexOf` on the relevant slice to handle potentially nested closing tags.
- * If the input string ends mid-block, the last open block is added and marked as partial.
- *
- * @param assistantMessage The raw string output from the assistant.
- * @returns An array of `AssistantMessageContent` objects, which can be `TextContent` or `ToolUse`.
- *          Blocks that were not fully closed by the end of the input string will have their `partial` flag set to `true`.
- */
-export function parseAssistantMessageV2(assistantMessage: string): AssistantMessageContent[] {
-	const contentBlocks: AssistantMessageContent[] = []
-	let currentTextContentStart = 0 // Index where the current text block started
-	let currentTextContent: TextContent | undefined = undefined
-	let currentToolUseStart = 0 // Index *after* the opening tag of the current tool use
-	let currentToolUse: ToolUse | undefined = undefined
-	let currentParamValueStart = 0 // Index *after* the opening tag of the current param
-	let currentParamName: ToolParamName | undefined = undefined
-
-	// Precompute tags for faster lookups
-	const toolUseOpenTags = new Map<string, ToolUseName>()
-	const toolParamOpenTags = new Map<string, ToolParamName>()
-	for (const name of toolUseNames) {
-		toolUseOpenTags.set(`<${name}>`, name)
-	}
-	for (const name of toolParamNames) {
-		toolParamOpenTags.set(`<${name}>`, name)
-	}
-
-	const len = assistantMessage.length
-	for (let i = 0; i < len; i++) {
-		const currentCharIndex = i
-
-		// --- State: Parsing a Tool Parameter ---
-		if (currentToolUse && currentParamName) {
-			const closeTag = `</${currentParamName}>`
-			// Check if the string *ending* at index `i` matches the closing tag
-			if (
-				currentCharIndex >= closeTag.length - 1 &&
-				assistantMessage.startsWith(
-					closeTag,
-					currentCharIndex - closeTag.length + 1, // Start checking from potential start of tag
-				)
-			) {
-				// Found the closing tag for the parameter
-				const value = assistantMessage
-					.slice(
-						currentParamValueStart, // Start after the opening tag
-						currentCharIndex - closeTag.length + 1, // End before the closing tag
-					)
-					.trim()
-				currentToolUse.params[currentParamName] = value
-				currentParamName = undefined // Go back to parsing tool content
-				// We don't continue loop here, need to check for tool close or other params at index i
-			} else {
-				continue // Still inside param value, move to next char
-			}
-		}
-
-		// --- State: Parsing a Tool Use (but not a specific parameter) ---
-		if (currentToolUse && !currentParamName) {
-			// Ensure we are not inside a parameter already
-			// Check if starting a new parameter
-			let startedNewParam = false
-			for (const [tag, paramName] of toolParamOpenTags.entries()) {
-				if (currentCharIndex >= tag.length - 1 && assistantMessage.startsWith(tag, currentCharIndex - tag.length + 1)) {
-					currentParamName = paramName
-					currentParamValueStart = currentCharIndex + 1 // Value starts after the tag
-					startedNewParam = true
-					break
-				}
-			}
-			if (startedNewParam) {
-				continue // Handled start of param, move to next char
-			}
-
-			// Check if closing the current tool use
-			const toolCloseTag = `</${currentToolUse.name}>`
-			if (
-				currentCharIndex >= toolCloseTag.length - 1 &&
-				assistantMessage.startsWith(toolCloseTag, currentCharIndex - toolCloseTag.length + 1)
-			) {
-				// End of the tool use found
-				// Special handling for content params *before* finalizing the tool
-				const toolContentSlice = assistantMessage.slice(
-					currentToolUseStart, // From after the tool opening tag
-					currentCharIndex - toolCloseTag.length + 1, // To before the tool closing tag
-				)
-
-				// Check if content parameter needs special handling (write_to_file/new_rule)
-				// This check is important if the closing </content> tag was missed by the parameter parsing logic
-				// (e.g., if content is empty or parsing logic prioritizes tool close)
-				const contentParamName: ToolParamName = "content"
-				if (
-					currentToolUse.name === "write_to_file" /* || currentToolUse.name === "new_rule" */ &&
-					toolContentSlice.includes(`<${contentParamName}>`)
-				) {
-					const contentStartTag = `<${contentParamName}>`
-					const contentEndTag = `</${contentParamName}>`
-					const contentStart = toolContentSlice.indexOf(contentStartTag)
-					// Use lastIndexOf for robustness against nested tags
-					const contentEnd = toolContentSlice.lastIndexOf(contentEndTag)
-
-					if (contentStart !== -1 && contentEnd !== -1 && contentEnd > contentStart) {
-						const contentValue = toolContentSlice.slice(contentStart + contentStartTag.length, contentEnd).trim()
-						currentToolUse.params[contentParamName] = contentValue
-					}
-				}
-
-				currentToolUse.partial = false // Mark as complete
-				contentBlocks.push(currentToolUse)
-				currentToolUse = undefined // Reset state
-				currentTextContentStart = currentCharIndex + 1 // Potential text starts after this tag
-				continue // Move to next char
-			}
-			// If not starting a param and not closing the tool, continue accumulating tool content implicitly
-			continue
-		}
-
-		// --- State: Parsing Text / Looking for Tool Start ---
-		if (!currentToolUse) {
-			// Check if starting a new tool use
-			let startedNewTool = false
-			for (const [tag, toolName] of toolUseOpenTags.entries()) {
-				if (currentCharIndex >= tag.length - 1 && assistantMessage.startsWith(tag, currentCharIndex - tag.length + 1)) {
-					// End current text block if one was active
-					if (currentTextContent) {
-						currentTextContent.content = assistantMessage
-							.slice(
-								currentTextContentStart, // From where text started
-								currentCharIndex - tag.length + 1, // To before the tool tag starts
-							)
-							.trim()
-						currentTextContent.partial = false // Ended because tool started
-						if (currentTextContent.content.length > 0) {
-							contentBlocks.push(currentTextContent)
-						}
-						currentTextContent = undefined
-					} else {
-						// Check for any text between the last block and this tag
-						const potentialText = assistantMessage
-							.slice(
-								currentTextContentStart, // From where text *might* have started
-								currentCharIndex - tag.length + 1, // To before the tool tag starts
-							)
-							.trim()
-						if (potentialText.length > 0) {
-							contentBlocks.push({
-								type: "text",
-								content: potentialText,
-								partial: false,
-							})
-						}
-					}
-
-					// Start the new tool use
-					currentToolUse = {
-						type: "tool_use",
-						name: toolName,
-						params: {},
-						partial: true, // Assume partial until closing tag is found
-					}
-					currentToolUseStart = currentCharIndex + 1 // Tool content starts after the opening tag
-					startedNewTool = true
-					break
-				}
-			}
-
-			if (startedNewTool) {
-				continue // Handled start of tool, move to next char
-			}
-
-			// If not starting a tool, it must be text content
-			if (!currentTextContent) {
-				// Start a new text block if we aren't already in one
-				currentTextContentStart = currentCharIndex // Text starts at the current character
-				// Check if the current char is the start of potential text *immediately* after a tag
-				// This needs the previous state - simpler to let slicing handle it later.
-				// Resetting start index accurately is key.
-				// It should be the index *after* the last processed tag.
-				// The logic managing currentTextContentStart after closing tags handles this.
-
-				currentTextContent = {
-					type: "text",
-					content: "", // Will be determined by slicing at the end or when a tool starts
-					partial: true,
-				}
-			}
-			// Continue accumulating text implicitly; content is extracted later.
-		}
-	} // End of loop
-
-	// --- Finalization after loop ---
-
-	// Finalize any open parameter within an open tool use
-	if (currentToolUse && currentParamName) {
-		currentToolUse.params[currentParamName] = assistantMessage
-			.slice(currentParamValueStart) // From param start to end of string
-			.trim()
-		// Tool use remains partial
-	}
-
-	// Finalize any open tool use (which might contain the finalized partial param)
-	if (currentToolUse) {
-		// Tool use is partial because the loop finished before its closing tag
-		contentBlocks.push(currentToolUse)
-	}
-	// Finalize any trailing text content
-	// Only possible if a tool use wasn't open at the very end
-	else if (currentTextContent) {
-		currentTextContent.content = assistantMessage
-			.slice(currentTextContentStart) // From text start to end of string
-			.trim()
-		// Text is partial because the loop finished
-		if (currentTextContent.content.length > 0) {
-			contentBlocks.push(currentTextContent)
-		}
+	// Note: it doesn't matter if check for currentToolUse or currentTextContent, only one of them will be defined since only one can be partial at a time
+	if (currentTextContent) {
+		// stream did not complete text content, add it as partial
+		contentBlocks.push(currentTextContent)
 	}
 
 	return contentBlocks

src/core/controller/account/accountLoginClicked.ts

@@ -2,7 +2,6 @@ import * as vscode from "vscode"
 import crypto from "crypto"
 import { Controller } from "../index"
 import { storeSecret } from "../../storage/state"
-import { EmptyRequest, String } from "../../../shared/proto/common"
 
 /**
  * Handles the user clicking the login link in the UI.
@@ -12,7 +11,7 @@ import { EmptyRequest, String } from "../../../shared/proto/common"
  * @param controller The controller instance.
  * @returns The login URL as a string.
  */
-export async function accountLoginClicked(controller: Controller, unused: EmptyRequest): Promise<String> {
+export async function accountLoginClicked(controller: Controller): Promise<String> {
 	// Generate nonce for state validation
 	const nonce = crypto.randomBytes(32).toString("hex")
 	await storeSecret(controller.context, "authNonce", nonce)
@@ -26,8 +25,6 @@ export async function accountLoginClicked(controller: Controller, unused: EmptyR
 	const authUrl = vscode.Uri.parse(
 		`https://app.cline.bot/auth?state=${encodeURIComponent(nonce)}&callback_url=${encodeURIComponent(`${uriScheme || "vscode"}://saoudrizwan.claude-dev/auth`)}`,
 	)
-	await vscode.env.openExternal(authUrl)
-	return {
-		value: authUrl.toString(),
-	}
+	vscode.env.openExternal(authUrl)
+	return authUrl.toString()
 }

src/core/controller/browser/updateBrowserSettings.ts

@@ -1,8 +1,8 @@
 import { UpdateBrowserSettingsRequest } from "../../../shared/proto/browser"
 import { Boolean } from "../../../shared/proto/common"
 import { Controller } from "../index"
-import { updateGlobalState, getGlobalState } from "../../storage/state"
-import { BrowserSettings as SharedBrowserSettings, DEFAULT_BROWSER_SETTINGS } from "../../../shared/BrowserSettings"
+import { updateGlobalState } from "../../storage/state"
+import { BrowserSettings as SharedBrowserSettings } from "../../../shared/BrowserSettings"
 
 /**
  * Update browser settings
@@ -12,39 +12,23 @@ import { BrowserSettings as SharedBrowserSettings, DEFAULT_BROWSER_SETTINGS } fr
  */
 export async function updateBrowserSettings(controller: Controller, request: UpdateBrowserSettingsRequest): Promise<Boolean> {
 	try {
-		// Get current browser settings to preserve fields not in the request
-		const currentSettings = (await getGlobalState(controller.context, "browserSettings")) as SharedBrowserSettings | undefined
-		const mergedWithDefaults = { ...DEFAULT_BROWSER_SETTINGS, ...currentSettings }
-
-		// Convert from protobuf format to shared format, merging with existing settings
-		const newBrowserSettings: SharedBrowserSettings = {
-			...mergedWithDefaults, // Start with existing settings (and defaults)
+		// Convert from protobuf format to shared format
+		const browserSettings: SharedBrowserSettings = {
 			viewport: {
-				// Apply updates from request
-				width: request.viewport?.width || mergedWithDefaults.viewport.width,
-				height: request.viewport?.height || mergedWithDefaults.viewport.height,
+				width: request.viewport?.width || 900,
+				height: request.viewport?.height || 600,
 			},
-			// Explicitly handle optional boolean and string fields from the request
-			remoteBrowserEnabled:
-				request.remoteBrowserEnabled === undefined
-					? mergedWithDefaults.remoteBrowserEnabled
-					: request.remoteBrowserEnabled,
-			remoteBrowserHost:
-				request.remoteBrowserHost === undefined ? mergedWithDefaults.remoteBrowserHost : request.remoteBrowserHost,
-			chromeExecutablePath:
-				// If chromeExecutablePath is explicitly in the request (even as ""), use it.
-				// Otherwise, fall back to mergedWithDefaults.
-				"chromeExecutablePath" in request ? request.chromeExecutablePath : mergedWithDefaults.chromeExecutablePath,
-			disableToolUse: request.disableToolUse === undefined ? mergedWithDefaults.disableToolUse : request.disableToolUse,
+			remoteBrowserEnabled: request.remoteBrowserEnabled || false,
+			remoteBrowserHost: request.remoteBrowserHost || undefined,
 		}
 
 		// Update global state with new settings
-		await updateGlobalState(controller.context, "browserSettings", newBrowserSettings)
+		await updateGlobalState(controller.context, "browserSettings", browserSettings)
 
 		// Update task browser settings if task exists
 		if (controller.task) {
-			controller.task.browserSettings = newBrowserSettings
-			controller.task.browserSession.browserSettings = newBrowserSettings
+			controller.task.browserSettings = browserSettings
+			controller.task.browserSession.browserSettings = browserSettings
 		}
 
 		// Post updated state to webview

src/core/controller/grpc-service-config.ts

@@ -12,7 +12,6 @@ import { handleStateServiceRequest, handleStateServiceStreamingRequest } from ".
 import { handleTaskServiceRequest, handleTaskServiceStreamingRequest } from "./task/index"
 import { handleWebServiceRequest, handleWebServiceStreamingRequest } from "./web/index"
 import { handleModelsServiceRequest, handleModelsServiceStreamingRequest } from "./models/index"
-import { handleSlashServiceRequest, handleSlashServiceStreamingRequest } from "./slash/index"
 
 /**
  * Configuration for a service handler
@@ -68,8 +67,4 @@ export const serviceHandlers: Record<string, ServiceHandlerConfig> = {
 		requestHandler: handleModelsServiceRequest,
 		streamingHandler: handleModelsServiceStreamingRequest,
 	},
-	"cline.SlashService": {
-		requestHandler: handleSlashServiceRequest,
-		streamingHandler: handleSlashServiceStreamingRequest,
-	},
 }

src/core/controller/index.ts

@@ -66,7 +66,7 @@ export class Controller {
 	workspaceTracker: WorkspaceTracker
 	mcpHub: McpHub
 	accountService: ClineAccountService
-	private latestAnnouncementId = "may-09-2025_17:11:00" // update to some unique identifier when we add a new announcement
+	private latestAnnouncementId = "may-02-2025_16:27:00" // update to some unique identifier when we add a new announcement
 
 	constructor(
 		readonly context: vscode.ExtensionContext,
@@ -249,7 +249,7 @@ export class Controller {
 				// If user already opted in to telemetry, enable telemetry service
 				this.getStateToPostToWebview().then((state) => {
 					const { telemetrySetting } = state
-					const isOptedIn = telemetrySetting !== "disabled"
+					const isOptedIn = telemetrySetting === "enabled"
 					telemetryService.updateTelemetryState(isOptedIn)
 				})
 				break
@@ -271,6 +271,12 @@ export class Controller {
 				// initializing new instance of Cline will make sure that any agentically running promises in old instance don't affect our new task. this essentially creates a fresh slate for the new task
 				await this.initTask(message.text, message.images)
 				break
+			case "condense":
+				this.task?.handleWebviewAskResponse("yesButtonClicked")
+				break
+			case "reportBug":
+				this.task?.handleWebviewAskResponse("yesButtonClicked")
+				break
 			case "apiConfiguration":
 				if (message.apiConfiguration) {
 					await updateApiConfiguration(this.context, message.apiConfiguration)
@@ -328,6 +334,9 @@ export class Controller {
 			case "resetState":
 				await this.resetState()
 				break
+			case "refreshRequestyModels":
+				await this.refreshRequestyModels()
+				break
 			case "refreshClineRules":
 				await refreshClineRulesToggles(this.context, cwd)
 				await refreshExternalRulesToggles(this.context, cwd)
@@ -373,6 +382,19 @@ export class Controller {
 				await this.fetchMcpMarketplace(message.bool)
 				break
 			}
+			case "downloadMcp": {
+				if (message.mcpId) {
+					// 1. Toggle to act mode if we are in plan mode
+					const { chatSettings } = await this.getStateToPostToWebview()
+					if (chatSettings.mode === "plan") {
+						await this.togglePlanActModeWithChatSettings({ mode: "act" })
+					}
+
+					// 2. download MCP
+					await this.downloadMcp(message.mcpId)
+				}
+				break
+			}
 			case "silentlyRefreshMcpMarketplace": {
 				await this.silentlyRefreshMcpMarketplace()
 				break
@@ -590,6 +612,27 @@ export class Controller {
 				this.postMessageToWebview({ type: "relinquishControl" })
 				break
 			}
+			case "toggleFavoriteModel": {
+				if (message.modelId) {
+					const { apiConfiguration } = await getAllExtensionState(this.context)
+					const favoritedModelIds = apiConfiguration.favoritedModelIds || []
+
+					// Toggle favorite status
+					const updatedFavorites = favoritedModelIds.includes(message.modelId)
+						? favoritedModelIds.filter((id) => id !== message.modelId)
+						: [...favoritedModelIds, message.modelId]
+
+					await updateGlobalState(this.context, "favoritedModelIds", updatedFavorites)
+
+					// Capture telemetry for model favorite toggle
+					const isFavorited = !favoritedModelIds.includes(message.modelId)
+					telemetryService.captureModelFavoritesUsage(message.modelId, isFavorited)
+
+					// Post state to webview without changing any other configuration
+					await this.postStateToWebview()
+				}
+				break
+			}
 			case "grpc_request": {
 				if (message.grpc_request) {
 					await handleGrpcRequest(this, message.grpc_request)
@@ -633,7 +676,7 @@ export class Controller {
 
 	async updateTelemetrySetting(telemetrySetting: TelemetrySetting) {
 		await updateGlobalState(this.context, "telemetrySetting", telemetrySetting)
-		const isOptedIn = telemetrySetting !== "disabled"
+		const isOptedIn = telemetrySetting === "enabled"
 		telemetryService.updateTelemetryState(isOptedIn)
 	}
 
@@ -713,7 +756,6 @@ export class Controller {
 					break
 				case "litellm":
 					await updateGlobalState(this.context, "previousModeModelId", apiConfiguration.liteLlmModelId)
-					await updateGlobalState(this.context, "previousModeModelInfo", apiConfiguration.liteLlmModelInfo)
 					break
 				case "requesty":
 					await updateGlobalState(this.context, "previousModeModelId", apiConfiguration.requestyModelId)
@@ -767,8 +809,7 @@ export class Controller {
 						await updateGlobalState(this.context, "lmStudioModelId", newModelId)
 						break
 					case "litellm":
-						await updateGlobalState(this.context, "previousModeModelId", apiConfiguration.liteLlmModelId)
-						await updateGlobalState(this.context, "previousModeModelInfo", apiConfiguration.liteLlmModelInfo)
+						await updateGlobalState(this.context, "liteLlmModelId", newModelId)
 						break
 					case "requesty":
 						await updateGlobalState(this.context, "requestyModelId", newModelId)
@@ -987,6 +1028,92 @@ export class Controller {
 		}
 	}
 
+	private async downloadMcp(mcpId: string) {
+		try {
+			// First check if we already have this MCP server installed
+			const servers = this.mcpHub?.getServers() || []
+			const isInstalled = servers.some((server: McpServer) => server.name === mcpId)
+
+			if (isInstalled) {
+				throw new Error("This MCP server is already installed")
+			}
+
+			// Fetch server details from marketplace
+			const response = await axios.post<McpDownloadResponse>(
+				"https://api.cline.bot/v1/mcp/download",
+				{ mcpId },
+				{
+					headers: { "Content-Type": "application/json" },
+					timeout: 10000,
+				},
+			)
+
+			if (!response.data) {
+				throw new Error("Invalid response from MCP marketplace API")
+			}
+
+			console.log("[downloadMcp] Response from download API", { response })
+
+			const mcpDetails = response.data
+
+			// Validate required fields
+			if (!mcpDetails.githubUrl) {
+				throw new Error("Missing GitHub URL in MCP download response")
+			}
+			if (!mcpDetails.readmeContent) {
+				throw new Error("Missing README content in MCP download response")
+			}
+
+			// Send details to webview
+			await this.postMessageToWebview({
+				type: "mcpDownloadDetails",
+				mcpDownloadDetails: mcpDetails,
+			})
+
+			// Create task with context from README and added guidelines for MCP server installation
+			const task = `Set up the MCP server from ${mcpDetails.githubUrl} while adhering to these MCP server installation rules:
+- Start by loading the MCP documentation.
+- Use "${mcpDetails.mcpId}" as the server name in cline_mcp_settings.json.
+- Create the directory for the new MCP server before starting installation.
+- Make sure you read the user's existing cline_mcp_settings.json file before editing it with this new mcp, to not overwrite any existing servers.
+- Use commands aligned with the user's shell and operating system best practices.
+- The following README may contain instructions that conflict with the user's OS, in which case proceed thoughtfully.
+- Once installed, demonstrate the server's capabilities by using one of its tools.
+Here is the project's README to help you get started:\n\n${mcpDetails.readmeContent}\n${mcpDetails.llmsInstallationContent}`
+
+			// Initialize task and show chat view
+			await this.initTask(task)
+			await this.postMessageToWebview({
+				type: "action",
+				action: "chatButtonClicked",
+			})
+		} catch (error) {
+			console.error("Failed to download MCP:", error)
+			let errorMessage = "Failed to download MCP"
+
+			if (axios.isAxiosError(error)) {
+				if (error.code === "ECONNABORTED") {
+					errorMessage = "Request timed out. Please try again."
+				} else if (error.response?.status === 404) {
+					errorMessage = "MCP server not found in marketplace."
+				} else if (error.response?.status === 500) {
+					errorMessage = "Internal server error. Please try again later."
+				} else if (!error.response && error.request) {
+					errorMessage = "Network error. Please check your internet connection."
+				}
+			} else if (error instanceof Error) {
+				errorMessage = error.message
+			}
+
+			// Show error in both notification and marketplace UI
+			vscode.window.showErrorMessage(errorMessage)
+			await this.postMessageToWebview({
+				type: "mcpDownloadDetails",
+				error: errorMessage,
+			})
+		}
+	}
+
 	// OpenRouter
 
 	async handleOpenRouterCallback(code: string) {
@@ -1022,7 +1149,6 @@ export class Controller {
 		return cacheDir
 	}
 
-	// Read OpenRouter models from disk cache
 	async readOpenRouterModels(): Promise<Record<string, ModelInfo> | undefined> {
 		const openRouterModelsFilePath = path.join(await this.ensureCacheDirectoryExists(), GlobalFileNames.openRouterModels)
 		const fileExists = await fileExistsAtPath(openRouterModelsFilePath)
@@ -1033,6 +1159,51 @@ export class Controller {
 		return undefined
 	}
 
+	async refreshRequestyModels() {
+		const parsePrice = (price: any) => {
+			if (price) {
+				return parseFloat(price) * 1_000_000
+			}
+			return undefined
+		}
+
+		let models: Record<string, ModelInfo> = {}
+		try {
+			const apiKey = await getSecret(this.context, "requestyApiKey")
+			const headers = {
+				Authorization: `Bearer ${apiKey}`,
+			}
+			const response = await axios.get("https://router.requesty.ai/v1/models", { headers })
+			if (response.data?.data) {
+				for (const model of response.data.data) {
+					const modelInfo: ModelInfo = {
+						maxTokens: model.max_output_tokens || undefined,
+						contextWindow: model.context_window,
+						supportsImages: model.supports_vision || undefined,
+						supportsPromptCache: model.supports_caching || undefined,
+						inputPrice: parsePrice(model.input_price),
+						outputPrice: parsePrice(model.output_price),
+						cacheWritesPrice: parsePrice(model.caching_price),
+						cacheReadsPrice: parsePrice(model.cached_price),
+						description: model.description,
+					}
+					models[model.id] = modelInfo
+				}
+				console.log("Requesty models fetched", models)
+			} else {
+				console.error("Invalid response from Requesty API")
+			}
+		} catch (error) {
+			console.error("Error fetching Requesty models:", error)
+		}
+
+		await this.postMessageToWebview({
+			type: "requestyModels",
+			requestyModels: models,
+		})
+		return models
+	}
+
 	// Context menus and code actions
 
 	getFileMentionFromPath(filePath: string) {

src/core/controller/mcp/downloadMcp.ts

@@ -1,114 +0,0 @@
-import { Controller } from ".."
-import { Empty, StringRequest } from "../../../shared/proto/common"
-import { McpServer, McpDownloadResponse } from "@shared/mcp"
-import axios from "axios"
-import * as vscode from "vscode"
-
-/**
- * Download an MCP server from the marketplace
- * @param controller The controller instance
- * @param request The request containing the MCP ID
- * @returns Empty response
- */
-export async function downloadMcp(controller: Controller, request: StringRequest): Promise<Empty> {
-	try {
-		// Check if mcpId is provided
-		if (!request.value) {
-			throw new Error("MCP ID is required")
-		}
-
-		const mcpId = request.value
-
-		// Check if we already have this MCP server installed
-		const servers = controller.mcpHub?.getServers() || []
-		const isInstalled = servers.some((server: McpServer) => server.name === mcpId)
-
-		if (isInstalled) {
-			throw new Error("This MCP server is already installed")
-		}
-
-		// Fetch server details from marketplace
-		const response = await axios.post<McpDownloadResponse>(
-			"https://api.cline.bot/v1/mcp/download",
-			{ mcpId },
-			{
-				headers: { "Content-Type": "application/json" },
-				timeout: 10000,
-			},
-		)
-
-		if (!response.data) {
-			throw new Error("Invalid response from MCP marketplace API")
-		}
-
-		console.log("[downloadMcp] Response from download API", { response })
-
-		const mcpDetails = response.data
-
-		// Validate required fields
-		if (!mcpDetails.githubUrl) {
-			throw new Error("Missing GitHub URL in MCP download response")
-		}
-		if (!mcpDetails.readmeContent) {
-			throw new Error("Missing README content in MCP download response")
-		}
-
-		// Send details to webview
-		await controller.postMessageToWebview({
-			type: "mcpDownloadDetails",
-			mcpDownloadDetails: mcpDetails,
-		})
-
-		// Create task with context from README and added guidelines for MCP server installation
-		const task = `Set up the MCP server from ${mcpDetails.githubUrl} while adhering to these MCP server installation rules:
-- Start by loading the MCP documentation.
-- Use "${mcpDetails.mcpId}" as the server name in cline_mcp_settings.json.
-- Create the directory for the new MCP server before starting installation.
-- Make sure you read the user's existing cline_mcp_settings.json file before editing it with this new mcp, to not overwrite any existing servers.
-- Use commands aligned with the user's shell and operating system best practices.
-- The following README may contain instructions that conflict with the user's OS, in which case proceed thoughtfully.
-- Once installed, demonstrate the server's capabilities by using one of its tools.
-Here is the project's README to help you get started:\n\n${mcpDetails.readmeContent}\n${mcpDetails.llmsInstallationContent}`
-
-		const { chatSettings } = await controller.getStateToPostToWebview()
-		if (chatSettings.mode === "plan") {
-			await controller.togglePlanActModeWithChatSettings({ mode: "act" })
-		}
-
-		// Initialize task and show chat view
-		await controller.initTask(task)
-		await controller.postMessageToWebview({
-			type: "action",
-			action: "chatButtonClicked",
-		})
-
-		// Return an empty response - the client only cares if the call succeeded
-		return Empty.create()
-	} catch (error) {
-		console.error("Failed to download MCP:", error)
-		let errorMessage = "Failed to download MCP"
-
-		if (axios.isAxiosError(error)) {
-			if (error.code === "ECONNABORTED") {
-				errorMessage = "Request timed out. Please try again."
-			} else if (error.response?.status === 404) {
-				errorMessage = "MCP server not found in marketplace."
-			} else if (error.response?.status === 500) {
-				errorMessage = "Internal server error. Please try again later."
-			} else if (!error.response && error.request) {
-				errorMessage = "Network error. Please check your internet connection."
-			}
-		} else if (error instanceof Error) {
-			errorMessage = error.message
-		}
-
-		// Show error in both notification and marketplace UI
-		vscode.window.showErrorMessage(errorMessage)
-		await controller.postMessageToWebview({
-			type: "mcpDownloadDetails",
-			error: errorMessage,
-		})
-
-		throw error
-	}
-}

src/core/controller/mcp/methods.ts

@@ -4,7 +4,6 @@
 // Import all method implementations
 import { registerMethod } from "./index"
 import { addRemoteMcpServer } from "./addRemoteMcpServer"
-import { downloadMcp } from "./downloadMcp"
 import { toggleMcpServer } from "./toggleMcpServer"
 import { updateMcpTimeout } from "./updateMcpTimeout"
 
@@ -12,7 +11,6 @@ import { updateMcpTimeout } from "./updateMcpTimeout"
 export function registerAllMethods(): void {
 	// Register each method with the registry
 	registerMethod("addRemoteMcpServer", addRemoteMcpServer)
-	registerMethod("downloadMcp", downloadMcp)
 	registerMethod("toggleMcpServer", toggleMcpServer)
 	registerMethod("updateMcpTimeout", updateMcpTimeout)
 }

src/core/controller/models/methods.ts

@@ -8,7 +8,6 @@ import { getOllamaModels } from "./getOllamaModels"
 import { getVsCodeLmModels } from "./getVsCodeLmModels"
 import { refreshOpenAiModels } from "./refreshOpenAiModels"
 import { refreshOpenRouterModels } from "./refreshOpenRouterModels"
-import { refreshRequestyModels } from "./refreshRequestyModels"
 
 // Register all models service methods
 export function registerAllMethods(): void {
@@ -18,5 +17,4 @@ export function registerAllMethods(): void {
 	registerMethod("getVsCodeLmModels", getVsCodeLmModels)
 	registerMethod("refreshOpenAiModels", refreshOpenAiModels)
 	registerMethod("refreshOpenRouterModels", refreshOpenRouterModels)
-	registerMethod("refreshRequestyModels", refreshRequestyModels)
 }

src/core/controller/models/refreshOpenAiModels.ts

@@ -29,6 +29,9 @@ export async function refreshOpenAiModels(controller: Controller, request: OpenA
 		const modelsArray = response.data?.data?.map((model: any) => model.id) || []
 		const models = [...new Set<string>(modelsArray)]
 
+		// Send models to webview
+		controller.postMessageToWebview({ type: "openAiModels", openAiModels: models })
+
 		return StringArray.create({ values: models })
 	} catch (error) {
 		console.error("Error fetching OpenAI models:", error)

src/core/controller/models/refreshOpenRouterModels.ts

@@ -1,6 +1,6 @@
 import { Controller } from ".."
 import { EmptyRequest } from "../../../shared/proto/common"
-import { OpenRouterCompatibleModelInfo, OpenRouterModelInfo } from "../../../shared/proto/models"
+import { OpenRouterModels, OpenRouterModelInfo } from "../../../shared/proto/models"
 import axios from "axios"
 import path from "path"
 import fs from "fs/promises"
@@ -13,10 +13,7 @@ import { GlobalFileNames } from "@core/storage/disk"
  * @param request Empty request object
  * @returns Response containing the OpenRouter models
  */
-export async function refreshOpenRouterModels(
-	controller: Controller,
-	request: EmptyRequest,
-): Promise<OpenRouterCompatibleModelInfo> {
+export async function refreshOpenRouterModels(controller: Controller, request: EmptyRequest): Promise<OpenRouterModels> {
 	const openRouterModelsFilePath = path.join(await ensureCacheDirectoryExists(controller), GlobalFileNames.openRouterModels)
 
 	let models: Record<string, Partial<OpenRouterModelInfo>> = {}
@@ -144,7 +141,13 @@ export async function refreshOpenRouterModels(
 		}
 	}
 
-	return OpenRouterCompatibleModelInfo.create({ models: typedModels })
+	// Send models to webview
+	await controller.postMessageToWebview({
+		type: "openRouterModels",
+		openRouterModels: typedModels,
+	})
+
+	return OpenRouterModels.create({ models: typedModels })
 }
 
 /**

src/core/controller/models/refreshRequestyModels.ts

@@ -1,55 +0,0 @@
-import { Controller } from ".."
-import { EmptyRequest } from "../../../shared/proto/common"
-import { OpenRouterCompatibleModelInfo, OpenRouterModelInfo } from "../../../shared/proto/models"
-import axios from "axios"
-import { getSecret } from "@core/storage/state"
-
-/**
- * Refreshes the Requesty models and returns the updated model list
- * @param controller The controller instance
- * @param request Empty request object
- * @returns Response containing the Requesty models
- */
-export async function refreshRequestyModels(
-	controller: Controller,
-	request: EmptyRequest,
-): Promise<OpenRouterCompatibleModelInfo> {
-	const parsePrice = (price: any) => {
-		if (price) {
-			return parseFloat(price) * 1_000_000
-		}
-		return undefined
-	}
-
-	let models: Record<string, OpenRouterModelInfo> = {}
-	try {
-		const apiKey = await getSecret(controller.context, "requestyApiKey")
-		const headers = {
-			Authorization: `Bearer ${apiKey}`,
-		}
-		const response = await axios.get("https://router.requesty.ai/v1/models", { headers })
-		if (response.data?.data) {
-			for (const model of response.data.data) {
-				const modelInfo: OpenRouterModelInfo = {
-					maxTokens: model.max_output_tokens || undefined,
-					contextWindow: model.context_window,
-					supportsImages: model.supports_vision || undefined,
-					supportsPromptCache: model.supports_caching || undefined,
-					inputPrice: parsePrice(model.input_price) || 0,
-					outputPrice: parsePrice(model.output_price) || 0,
-					cacheWritesPrice: parsePrice(model.caching_price) || 0,
-					cacheReadsPrice: parsePrice(model.cached_price) || 0,
-					description: model.description,
-				}
-				models[model.id] = modelInfo
-			}
-			console.log("Requesty models fetched", models)
-		} else {
-			console.error("Invalid response from Requesty API")
-		}
-	} catch (error) {
-		console.error("Error fetching Requesty models:", error)
-	}
-
-	return OpenRouterCompatibleModelInfo.create({ models })
-}

src/core/controller/slash/condense.ts

@@ -1,10 +0,0 @@
-import { Controller } from ".."
-import { StringRequest, Empty } from "../../../shared/proto/common"
-
-/**
- * Command slash command logic
- */
-export async function condense(controller: Controller, request: StringRequest): Promise<Empty> {
-	await controller.task?.handleWebviewAskResponse("yesButtonClicked")
-	return Empty.create()
-}

src/core/controller/slash/index.ts

@@ -1,22 +0,0 @@
-// AUTO-GENERATED FILE - DO NOT MODIFY DIRECTLY
-// Generated by proto/build-proto.js
-
-import { createServiceRegistry, ServiceMethodHandler, StreamingMethodHandler } from "../grpc-service"
-import { StreamingResponseHandler } from "../grpc-handler"
-import { registerAllMethods } from "./methods"
-
-// Create slash service registry
-const slashService = createServiceRegistry("slash")
-
-// Export the method handler types and registration function
-export type SlashMethodHandler = ServiceMethodHandler
-export type SlashStreamingMethodHandler = StreamingMethodHandler
-export const registerMethod = slashService.registerMethod
-
-// Export the request handlers
-export const handleSlashServiceRequest = slashService.handleRequest
-export const handleSlashServiceStreamingRequest = slashService.handleStreamingRequest
-export const isStreamingMethod = slashService.isStreamingMethod
-
-// Register all slash methods
-registerAllMethods()

src/core/controller/slash/methods.ts

@@ -1,14 +0,0 @@
-// AUTO-GENERATED FILE - DO NOT MODIFY DIRECTLY
-// Generated by proto/build-proto.js
-
-// Import all method implementations
-import { registerMethod } from "./index"
-import { condense } from "./condense"
-import { reportBug } from "./reportBug"
-
-// Register all slash service methods
-export function registerAllMethods(): void {
-	// Register each method with the registry
-	registerMethod("condense", condense)
-	registerMethod("reportBug", reportBug)
-}

src/core/controller/slash/reportBug.ts

@@ -1,10 +0,0 @@
-import { Controller } from ".."
-import { StringRequest, Empty } from "../../../shared/proto/common"
-
-/**
- * Report bug slash command logic
- */
-export async function reportBug(controller: Controller, request: StringRequest): Promise<Empty> {
-	await controller.task?.handleWebviewAskResponse("yesButtonClicked")
-	return Empty.create()
-}

src/core/controller/state/methods.ts

@@ -5,7 +5,6 @@
 import { registerMethod } from "./index"
 import { getLatestState } from "./getLatestState"
 import { subscribeToState } from "./subscribeToState"
-import { toggleFavoriteModel } from "./toggleFavoriteModel"
 
 // Streaming methods for this service
 export const streamingMethods = ["subscribeToState"]
@@ -15,5 +14,4 @@ export function registerAllMethods(): void {
 	// Register each method with the registry
 	registerMethod("getLatestState", getLatestState)
 	registerMethod("subscribeToState", subscribeToState, { isStreaming: true })
-	registerMethod("toggleFavoriteModel", toggleFavoriteModel)
 }

src/core/controller/state/toggleFavoriteModel.ts

@@ -1,46 +0,0 @@
-import { telemetryService } from "@/services/posthog/telemetry/TelemetryService"
-import { Controller } from ".."
-import { Empty, StringRequest } from "../../../shared/proto/common"
-import { updateGlobalState } from "@/core/storage/state"
-
-/**
- * Toggles a model's favorite status
- * @param controller The controller instance
- * @param request The request containing the model ID to toggle
- * @returns An empty response
- */
-export async function toggleFavoriteModel(controller: Controller, request: StringRequest): Promise<Empty> {
-	try {
-		if (!request.value) {
-			throw new Error("Model ID is required")
-		}
-
-		const modelId = request.value
-		const { apiConfiguration } = await controller.getStateToPostToWebview()
-
-		if (!apiConfiguration) {
-			throw new Error("API configuration not found")
-		}
-
-		const favoritedModelIds = apiConfiguration.favoritedModelIds || []
-
-		// Toggle favorite status
-		const updatedFavorites = favoritedModelIds.includes(modelId)
-			? favoritedModelIds.filter((id) => id !== modelId)
-			: [...favoritedModelIds, modelId]
-
-		await updateGlobalState(controller.context, "favoritedModelIds", updatedFavorites)
-
-		// Capture telemetry for model favorite toggle
-		const isFavorited = !favoritedModelIds.includes(modelId)
-		telemetryService.captureModelFavoritesUsage(modelId, isFavorited)
-
-		// Post state to webview without changing any other configuration
-		await controller.postStateToWebview()
-
-		return Empty.create()
-	} catch (error) {
-		console.error(`Failed to toggle favorite status for model ${request.value}:`, error)
-		throw error
-	}
-}

src/core/storage/state-keys.ts

@@ -11,7 +11,6 @@ export type SecretKey =
 	| "deepSeekApiKey"
 	| "requestyApiKey"
 	| "togetherApiKey"
-	| "fireworksApiKey"
 	| "qwenApiKey"
 	| "doubaoApiKey"
 	| "mistralApiKey"
@@ -68,11 +67,7 @@ export type GlobalStateKey =
 	| "previousModeModelInfo"
 	| "liteLlmBaseUrl"
 	| "liteLlmModelId"
-	| "liteLlmModelInfo"
 	| "liteLlmUsePromptCache"
-	| "fireworksModelId"
-	| "fireworksModelMaxCompletionTokens"
-	| "fireworksModelMaxTokens"
 	| "qwenApiLine"
 	| "requestyModelId"
 	| "requestyModelInfo"

src/core/storage/state.ts

@@ -107,12 +107,7 @@ export async function getAllExtensionState(context: vscode.ExtensionContext) {
 		vsCodeLmModelSelector,
 		liteLlmBaseUrl,
 		liteLlmModelId,
-		liteLlmModelInfo,
 		liteLlmUsePromptCache,
-		fireworksApiKey,
-		fireworksModelId,
-		fireworksModelMaxCompletionTokens,
-		fireworksModelMaxTokens,
 		userInfo,
 		previousModeApiProvider,
 		previousModeModelId,
@@ -191,12 +186,7 @@ export async function getAllExtensionState(context: vscode.ExtensionContext) {
 		getGlobalState(context, "vsCodeLmModelSelector") as Promise<vscode.LanguageModelChatSelector | undefined>,
 		getGlobalState(context, "liteLlmBaseUrl") as Promise<string | undefined>,
 		getGlobalState(context, "liteLlmModelId") as Promise<string | undefined>,
-		getGlobalState(context, "liteLlmModelInfo") as Promise<ModelInfo | undefined>,
 		getGlobalState(context, "liteLlmUsePromptCache") as Promise<boolean | undefined>,
-		getSecret(context, "fireworksApiKey") as Promise<string | undefined>,
-		getGlobalState(context, "fireworksModelId") as Promise<string | undefined>,
-		getGlobalState(context, "fireworksModelMaxCompletionTokens") as Promise<number | undefined>,
-		getGlobalState(context, "fireworksModelMaxTokens") as Promise<number | undefined>,
 		getGlobalState(context, "userInfo") as Promise<UserInfo | undefined>,
 		getGlobalState(context, "previousModeApiProvider") as Promise<ApiProvider | undefined>,
 		getGlobalState(context, "previousModeModelId") as Promise<string | undefined>,
@@ -314,13 +304,8 @@ export async function getAllExtensionState(context: vscode.ExtensionContext) {
 			reasoningEffort,
 			liteLlmBaseUrl,
 			liteLlmModelId,
-			liteLlmModelInfo,
 			liteLlmApiKey,
 			liteLlmUsePromptCache,
-			fireworksApiKey,
-			fireworksModelId,
-			fireworksModelMaxCompletionTokens,
-			fireworksModelMaxTokens,
 			asksageApiKey,
 			asksageApiUrl,
 			xaiApiKey,
@@ -401,7 +386,6 @@ export async function updateApiConfiguration(context: vscode.ExtensionContext, a
 		vsCodeLmModelSelector,
 		liteLlmBaseUrl,
 		liteLlmModelId,
-		liteLlmModelInfo,
 		liteLlmApiKey,
 		liteLlmUsePromptCache,
 		qwenApiLine,
@@ -460,7 +444,6 @@ export async function updateApiConfiguration(context: vscode.ExtensionContext, a
 	await updateGlobalState(context, "vsCodeLmModelSelector", vsCodeLmModelSelector)
 	await updateGlobalState(context, "liteLlmBaseUrl", liteLlmBaseUrl)
 	await updateGlobalState(context, "liteLlmModelId", liteLlmModelId)
-	await updateGlobalState(context, "liteLlmModelInfo", liteLlmModelInfo)
 	await updateGlobalState(context, "liteLlmUsePromptCache", liteLlmUsePromptCache)
 	await updateGlobalState(context, "qwenApiLine", qwenApiLine)
 	await updateGlobalState(context, "requestyModelId", requestyModelId)
@@ -497,7 +480,6 @@ export async function resetExtensionState(context: vscode.ExtensionContext) {
 		"mistralApiKey",
 		"clineApiKey",
 		"liteLlmApiKey",
-		"fireworksApiKey",
 		"asksageApiKey",
 		"xaiApiKey",
 		"sambanovaApiKey",

src/core/task/index.ts

@@ -60,7 +60,7 @@ import { fileExistsAtPath } from "@utils/fs"
 import { createAndOpenGitHubIssue } from "@utils/github-url-utils"
 import { arePathsEqual, getReadablePath, isLocatedInWorkspace } from "@utils/path"
 import { fixModelHtmlEscaping, removeInvalidChars } from "@utils/string"
-import { AssistantMessageContent, parseAssistantMessageV2, ToolParamName, ToolUseName } from "@core/assistant-message"
+import { AssistantMessageContent, parseAssistantMessage, ToolParamName, ToolUseName } from "@core/assistant-message"
 import { constructNewFileContent } from "@core/assistant-message/diff"
 import { ClineIgnoreController } from "@core/ignore/ClineIgnoreController"
 import { parseMentions } from "@core/mentions"
@@ -196,7 +196,9 @@ export class Task {
 		this.reinitExistingTaskFromId = reinitExistingTaskFromId
 		this.cancelTask = cancelTask
 		this.clineIgnoreController = new ClineIgnoreController(cwd)
-		// Initialization moved to startTask/resumeTaskFromHistory
+		this.clineIgnoreController.initialize().catch((error) => {
+			console.error("Failed to initialize ClineIgnoreController:", error)
+		})
 		this.terminalManager = new TerminalManager()
 		this.terminalManager.setShellIntegrationTimeout(shellIntegrationTimeout)
 		this.urlContentFetcher = new UrlContentFetcher(context)
@@ -855,12 +857,6 @@ export class Task {
 	// Task lifecycle
 
 	private async startTask(task?: string, images?: string[]): Promise<void> {
-		try {
-			await this.clineIgnoreController.initialize()
-		} catch (error) {
-			console.error("Failed to initialize ClineIgnoreController:", error)
-			// Optionally, inform the user or handle the error appropriately
-		}
 		// conversationHistory (for API) and clineMessages (for webview) need to be in sync
 		// if the extension process were killed, then on restart the clineMessages might not be empty, so we need to set it to [] when we create a new Cline client (otherwise webview would show stale messages from previous session)
 		this.clineMessages = []
@@ -883,12 +879,6 @@ export class Task {
 	}
 
 	private async resumeTaskFromHistory() {
-		try {
-			await this.clineIgnoreController.initialize()
-		} catch (error) {
-			console.error("Failed to initialize ClineIgnoreController:", error)
-			// Optionally, inform the user or handle the error appropriately
-		}
 		// UPDATE: we don't need this anymore since most tasks are now created with checkpoints enabled
 		// right now we let users init checkpoints for old tasks, assuming they're continuing them from the same workspace (which we never tied to tasks, so no way for us to know if it's opened in the right workspace)
 		// const doesShadowGitExist = await CheckpointTracker.doesShadowGitExist(this.taskId, this.controllerRef.deref())
@@ -1442,28 +1432,13 @@ export class Task {
 		return statusCode && !message.includes(statusCode.toString()) ? `${statusCode} - ${message}` : message
 	}
 
-	/**
-	 * Migrates the disableBrowserTool setting from VSCode configuration to browserSettings
-	 */
-	private async migrateDisableBrowserToolSetting(): Promise<void> {
-		const config = vscode.workspace.getConfiguration("cline")
-		const disableBrowserTool = vscode.workspace.getConfiguration("cline").get<boolean>("disableBrowserTool")
-
-		if (disableBrowserTool !== undefined) {
-			this.browserSettings.disableToolUse = disableBrowserTool
-			// Remove from VSCode configuration
-			await config.update("disableBrowserTool", undefined, true)
-		}
-	}
-
 	async *attemptApiRequest(previousApiReqIndex: number): ApiStream {
 		// Wait for MCP servers to be connected before generating system prompt
 		await pWaitFor(() => this.mcpHub.isConnecting !== true, { timeout: 10_000 }).catch(() => {
 			console.error("MCP servers failed to connect in time")
 		})
 
-		await this.migrateDisableBrowserToolSetting()
-		const disableBrowserTool = this.browserSettings.disableToolUse ?? false
+		const disableBrowserTool = vscode.workspace.getConfiguration("cline").get<boolean>("disableBrowserTool") ?? false
 		// cline browser tool uses image recognition for navigation (requires model image support).
 		const modelSupportsBrowserUse = this.api.getModel().info.supportsImages ?? false
 
@@ -3259,7 +3234,7 @@ export class Task {
 									await this.say("user_feedback", text ?? "", images)
 									pushToolResult(
 										formatResponse.toolResult(
-											`The user did not submit the bug, and provided feedback on the Github issue generated instead:\n<feedback>\n${text}\n</feedback>`,
+											`The user provided feedback on the Github issue generated:\n<feedback>\n${text}\n</feedback>`,
 											images,
 										),
 									)
@@ -3859,7 +3834,7 @@ export class Task {
 							assistantMessage += chunk.text
 							// parse raw assistant message into content blocks
 							const prevLength = this.assistantMessageContent.length
-							this.assistantMessageContent = parseAssistantMessageV2(assistantMessage)
+							this.assistantMessageContent = parseAssistantMessage(assistantMessage)
 							if (this.assistantMessageContent.length > prevLength) {
 								this.userMessageContentReady = false // new content we need to present, reset to false in case previous content set this to true
 							}