resolving merge conflicts

2026-09-16 21:01:52 +08:00 · 2025-03-04 15:12:46 -08:00
parent 53582d6149 ccba2ed8f8
commit 89ad9a241c
31 changed files with 577 additions and 205 deletions
@@ -0,0 +1,5 @@
+---
+"claude-dev": patch
+---
+
+Fixed problem with “win+shift+a” shortcut not working in Windows
@@ -0,0 +1,5 @@
+---
+"claude-dev": patch
+---
+
+Better Openrouter error typing and throwing more detailed messages
@@ -1,5 +0,0 @@
---
-"claude-dev": patch
---
-
-Add retries to Bedrock createMessage
@@ -0,0 +1,5 @@
+---
+"claude-dev": patch
+---
+
+Add webview to lint job
@@ -1,5 +0,0 @@
---
-"claude-dev": minor
---
-
-Update Claude 3.5 -> Claude 3.7 in error message/recommendation
@@ -1,5 +0,0 @@
---
-"claude-dev": patch
---
-
-Add support for AskSage as model provider.
@@ -1,5 +0,0 @@
---
-"claude-dev": patch
---
-
-Add timeout option to MCP servers
@@ -1,5 +0,0 @@
---
-"claude-dev": patch
---
-
-Add APAC Support for Cross-Region Inference Profiles
@@ -1 +1 @@
-*       @saoudrizwan @ocasta181 @NightTrek @pashpashpash
+*       @saoudrizwan @ocasta181 @NightTrek @pashpashpash @dcbartlett
@@ -1,5 +1,12 @@
 # Changelog

+## [3.5.1]
+
+-   Add timeout option to MCP servers
+-   Add Gemini Flash models to Vertex provider (thanks @jpaodev!)
+-   Add prompt caching support for AWS Bedrock provider (thanks @buger!)
+-   Add AskSage provider (thanks @swhite24!)
+
 ## [3.5.0]

 -   Add 'Enable extended thinking' option for Claude 3.7 Sonnet, with ability to set different budgets for Plan and Act modes
@@ -1,17 +1,18 @@
 {
 	"name": "claude-dev",
-	"version": "3.5.0",
+	"version": "3.5.1",
 	"lockfileVersion": 3,
 	"requires": true,
 	"packages": {
 		"": {
 			"name": "claude-dev",
-			"version": "3.5.0",
+			"version": "3.5.1",
 			"license": "Apache-2.0",
 			"dependencies": {
 				"@anthropic-ai/bedrock-sdk": "^0.12.4",
 				"@anthropic-ai/sdk": "^0.37.0",
 				"@anthropic-ai/vertex-sdk": "^0.6.4",
+				"@google-cloud/vertexai": "^1.9.3",
 				"@google/generative-ai": "^0.18.0",
 				"@mistralai/mistralai": "^1.5.0",
 				"@modelcontextprotocol/sdk": "^1.0.1",
@@ -36,6 +37,7 @@
 				"isbinaryfile": "^5.0.2",
 				"mammoth": "^1.8.0",
 				"monaco-vscode-textmate-theme-converter": "^0.1.7",
+				"ollama": "^0.5.13",
 				"open-graph-scraper": "^6.9.0",
 				"openai": "^4.83.0",
 				"os-name": "^6.0.0",
@@ -3787,6 +3789,18 @@
 			"resolved": "https://registry.npmjs.org/@firebase/webchannel-wrapper/-/webchannel-wrapper-1.0.3.tgz",
 			"integrity": "sha512-2xCRM9q9FlzGZCdgDMJwc0gyUkWFtkosy7Xxr6sFgQwn+wMNIWd7xIvYNauU1r64B5L5rsGKy/n9TKJ0aAFeqQ=="
 		},
+		"node_modules/@google-cloud/vertexai": {
+			"version": "1.9.3",
+			"resolved": "https://registry.npmjs.org/@google-cloud/vertexai/-/vertexai-1.9.3.tgz",
+			"integrity": "sha512-35o5tIEMLW3JeFJOaaMNR2e5sq+6rpnhrF97PuAxeOm0GlqVTESKhkGj7a5B5mmJSSSU3hUfIhcQCRRsw4Ipzg==",
+			"license": "Apache-2.0",
+			"dependencies": {
+				"google-auth-library": "^9.1.0"
+			},
+			"engines": {
+				"node": ">=18.0.0"
+			}
+		},
 		"node_modules/@google/generative-ai": {
 			"version": "0.18.0",
 			"resolved": "https://registry.npmjs.org/@google/generative-ai/-/generative-ai-0.18.0.tgz",
@@ -11051,6 +11065,15 @@
 				"url": "https://github.com/sponsors/ljharb"
 			}
 		},
+		"node_modules/ollama": {
+			"version": "0.5.13",
+			"resolved": "https://registry.npmjs.org/ollama/-/ollama-0.5.13.tgz",
+			"integrity": "sha512-qK3eE2GjMYjCiTknEJfAHjbUzUqgVtf9qtzjxWrkwBZgBG7kOB6Z4+Ov4fbvDjmKKHv+rpuTsWFg4jZvVjNBtQ==",
+			"license": "MIT",
+			"dependencies": {
+				"whatwg-fetch": "^3.6.20"
+			}
+		},
 		"node_modules/once": {
 			"version": "1.4.0",
 			"resolved": "https://registry.npmjs.org/once/-/once-1.4.0.tgz",
@@ -13401,6 +13424,12 @@
 				"node": ">=18"
 			}
 		},
+		"node_modules/whatwg-fetch": {
+			"version": "3.6.20",
+			"resolved": "https://registry.npmjs.org/whatwg-fetch/-/whatwg-fetch-3.6.20.tgz",
+			"integrity": "sha512-EqhiFU6daOA8kpjOWTL0olhVOF3i7OrFzSYiGsEMB8GcXS+RrzauAERX65xMeNWVqxA6HXH2m69Z9LaKKdisfg==",
+			"license": "MIT"
+		},
 		"node_modules/whatwg-mimetype": {
 			"version": "4.0.0",
 			"resolved": "https://registry.npmjs.org/whatwg-mimetype/-/whatwg-mimetype-4.0.0.tgz",
@@ -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": "3.5.0",
+	"version": "3.5.1",
 	"icon": "assets/icons/icon.png",
 	"galleryBanner": {
 		"color": "#617A91",
@@ -236,7 +236,7 @@
 		"watch-tests": "tsc -p . -w --outDir out",
 		"pretest": "npm run compile-tests && npm run compile && npm run lint",
 		"check-types": "tsc --noEmit",
-		"lint": "eslint src --ext ts",
+		"lint": "eslint src --ext ts && eslint webview-ui/src --ext ts",
 		"format": "prettier . --check",
 		"format:fix": "prettier . --write",
 		"test": "vscode-test",
@@ -275,6 +275,7 @@
 		"@anthropic-ai/bedrock-sdk": "^0.12.4",
 		"@anthropic-ai/sdk": "^0.37.0",
 		"@anthropic-ai/vertex-sdk": "^0.6.4",
+		"@google-cloud/vertexai": "^1.9.3",
 		"@google/generative-ai": "^0.18.0",
 		"@mistralai/mistralai": "^1.5.0",
 		"@modelcontextprotocol/sdk": "^1.0.1",
@@ -299,6 +300,7 @@
 		"isbinaryfile": "^5.0.2",
 		"mammoth": "^1.8.0",
 		"monaco-vscode-textmate-theme-converter": "^0.1.7",
+		"ollama": "^0.5.13",
 		"open-graph-scraper": "^6.9.0",
 		"openai": "^4.83.0",
 		"os-name": "^6.0.0",
@@ -19,11 +19,13 @@ export class AnthropicHandler implements ApiHandler {

 	@withRetry()
 	async *createMessage(systemPrompt: string, messages: Anthropic.Messages.MessageParam[]): ApiStream {
-		let budget_tokens = this.options.thinkingBudgetTokens || 0
-		const reasoningOn = budget_tokens !== 0 ? true : false
 		const model = this.getModel()
 		let stream: AnthropicStream<Anthropic.RawMessageStreamEvent>
 		const modelId = model.id
+
+		let budget_tokens = this.options.thinkingBudgetTokens || 0
+		const reasoningOn = modelId.includes("3-7") && budget_tokens !== 0 ? true : false
+
 		switch (modelId) {
 			// 'latest' alias does not support cache_control
 			case "claude-3-7-sonnet-20250219":
@@ -16,11 +16,12 @@ export class AwsBedrockHandler implements ApiHandler {

 	@withRetry()
 	async *createMessage(systemPrompt: string, messages: Anthropic.Messages.MessageParam[]): ApiStream {
-		let budget_tokens = this.options.thinkingBudgetTokens || 0
-		const reasoningOn = budget_tokens !== 0 ? true : false
 		// cross region inference requires prefixing the model id with the region
 		let modelId = await this.getModelId()

+		let budget_tokens = this.options.thinkingBudgetTokens || 0
+		const reasoningOn = modelId.includes("3-7") && budget_tokens !== 0 ? true : false
+
 		// Get model info and message indices for caching
 		const model = this.getModel()
 		const userMsgIndices = messages.reduce((acc, msg, index) => (msg.role === "user" ? [...acc, index] : acc), [] as number[])
@@ -33,11 +34,47 @@ export class AwsBedrockHandler implements ApiHandler {

 		const stream = await client.messages.create({
 			model: modelId,
-			max_tokens: this.getModel().info.maxTokens || 8192,
+			max_tokens: model.info.maxTokens || 8192,
 			thinking: reasoningOn ? { type: "enabled", budget_tokens: budget_tokens } : undefined,
 			temperature: reasoningOn ? undefined : 0,
-			system: systemPrompt,
-			messages,
+			system: [
+				{
+					text: systemPrompt,
+					type: "text",
+					...(this.options.awsBedrockUsePromptCache === true && {
+						cache_control: { type: "ephemeral" },
+					}),
+				},
+			],
+			messages: messages.map((message, index) => {
+				if (index === lastUserMsgIndex || index === secondLastMsgUserIndex) {
+					return {
+						...message,
+						content:
+							typeof message.content === "string"
+								? [
+										{
+											type: "text",
+											text: message.content,
+											...(this.options.awsBedrockUsePromptCache === true && {
+												cache_control: { type: "ephemeral" },
+											}),
+										},
+									]
+								: message.content.map((content, contentIndex) =>
+										contentIndex === message.content.length - 1
+											? {
+													...content,
+													...(this.options.awsBedrockUsePromptCache === true && {
+														cache_control: { type: "ephemeral" },
+													}),
+												}
+											: content,
+									),
+					}
+				}
+				return message
+			}),
 			stream: true,
 		})

@@ -1,40 +1,35 @@
 import { Anthropic } from "@anthropic-ai/sdk"
-import OpenAI from "openai"
+import { Message, Ollama } from "ollama"
 import { ApiHandler } from "../"
 import { ApiHandlerOptions, ModelInfo, openAiModelInfoSaneDefaults } from "../../shared/api"
-import { convertToOpenAiMessages } from "../transform/openai-format"
+import { convertToOllamaMessages } from "../transform/ollama-format"
 import { ApiStream } from "../transform/stream"

 export class OllamaHandler implements ApiHandler {
 	private options: ApiHandlerOptions
-	private client: OpenAI
+	private client: Ollama

 	constructor(options: ApiHandlerOptions) {
 		this.options = options
-		this.client = new OpenAI({
-			baseURL: (this.options.ollamaBaseUrl || "http://localhost:11434") + "/v1",
-			apiKey: "ollama",
-		})
+		this.client = new Ollama({ host: this.options.ollamaBaseUrl || "http://localhost:11434" })
 	}

 	async *createMessage(systemPrompt: string, messages: Anthropic.Messages.MessageParam[]): ApiStream {
-		const openAiMessages: OpenAI.Chat.ChatCompletionMessageParam[] = [
-			{ role: "system", content: systemPrompt },
-			...convertToOpenAiMessages(messages),
-		]
+		const ollamaMessages: Message[] = [{ role: "system", content: systemPrompt }, ...convertToOllamaMessages(messages)]

-		const stream = await this.client.chat.completions.create({
+		const stream = await this.client.chat({
 			model: this.getModel().id,
-			messages: openAiMessages,
-			temperature: 0,
+			messages: ollamaMessages,
 			stream: true,
+			options: {
+				num_ctx: Number(this.options.ollamaApiOptionsCtxNum) || 32768,
+			},
 		})
 		for await (const chunk of stream) {
-			const delta = chunk.choices[0]?.delta
-			if (delta?.content) {
+			if (typeof chunk.message.content === "string") {
 				yield {
 					type: "text",
-					text: delta.content,
+					text: chunk.message.content,
 				}
 			}
 		}
@@ -6,6 +6,7 @@ import { ApiHandler } from "../index"
 import { convertToOpenAiMessages } from "../transform/openai-format"
 import { ApiStream } from "../transform/stream"
 import { convertToR1Format } from "../transform/r1-format"
+import { ChatCompletionReasoningEffort } from "openai/resources/chat/completions.mjs"

 export class OpenAiHandler implements ApiHandler {
 	private options: ApiHandlerOptions
@@ -32,20 +33,30 @@ export class OpenAiHandler implements ApiHandler {
 	async *createMessage(systemPrompt: string, messages: Anthropic.Messages.MessageParam[]): ApiStream {
 		const modelId = this.options.openAiModelId ?? ""
 		const isDeepseekReasoner = modelId.includes("deepseek-reasoner")
+		const isO3Mini = modelId.includes("o3-mini")

 		let openAiMessages: OpenAI.Chat.ChatCompletionMessageParam[] = [
 			{ role: "system", content: systemPrompt },
 			...convertToOpenAiMessages(messages),
 		]
+		let temperature: number | undefined = 0
+		let reasoningEffort: ChatCompletionReasoningEffort | undefined = undefined

 		if (isDeepseekReasoner) {
 			openAiMessages = convertToR1Format([{ role: "user", content: systemPrompt }, ...messages])
 		}

+		if (isO3Mini) {
+			openAiMessages = [{ role: "developer", content: systemPrompt }, ...convertToOpenAiMessages(messages)]
+			temperature = undefined // does not support temperature
+			reasoningEffort = (this.options.o3MiniReasoningEffort as ChatCompletionReasoningEffort) || "medium"
+		}
+
 		const stream = await this.client.chat.completions.create({
 			model: modelId,
 			messages: openAiMessages,
-			temperature: 0,
+			temperature,
+			reasoning_effort: reasoningEffort,
 			stream: true,
 			stream_options: { include_usage: true },
 		})
@@ -7,6 +7,8 @@ import { ApiHandler } from "../"
 import { ApiHandlerOptions, ModelInfo, openRouterDefaultModelId, openRouterDefaultModelInfo } from "../../shared/api"
 import { streamOpenRouterFormatRequest } from "../transform/openrouter-stream"
 import { ApiStream } from "../transform/stream"
+import { convertToR1Format } from "../transform/r1-format"
+import { OpenRouterErrorResponse } from "./types"

 export class OpenRouterHandler implements ApiHandler {
 	private options: ApiHandlerOptions
@@ -0,0 +1,22 @@
+// For the following openrouter error type sources, see the docs here:
+// https://openrouter.ai/docs/api-reference/errors
+
+export type OpenRouterErrorResponse = {
+	error: {
+		message: string
+		code: number
+		metadata?: OpenRouterProviderErrorMetadata | OpenRouterModerationErrorMetadata | Record<string, unknown>
+	}
+}
+
+export type OpenRouterProviderErrorMetadata = {
+	provider_name: string // The name of the provider that encountered the error
+	raw: unknown // The raw error from the provider
+}
+
+export type OpenRouterModerationErrorMetadata = {
+	reasons: string[] // Why your input was flagged
+	flagged_input: string // The text segment that was flagged, limited to 100 characters. If the flagged input is longer than 100 characters, it will be truncated in the middle and replaced with ...
+	provider_name: string // The name of the provider that requested moderation
+	model_slug: string
+}
@@ -4,19 +4,25 @@ import { withRetry } from "../retry"
 import { ApiHandler } from "../"
 import { ApiHandlerOptions, ModelInfo, vertexDefaultModelId, VertexModelId, vertexModels } from "../../shared/api"
 import { ApiStream } from "../transform/stream"
+import { VertexAI } from "@google-cloud/vertexai"

 // https://docs.anthropic.com/en/api/claude-on-vertex-ai
 export class VertexHandler implements ApiHandler {
 	private options: ApiHandlerOptions
-	private client: AnthropicVertex
+	private clientAnthropic: AnthropicVertex
+	private clientVertex: VertexAI

 	constructor(options: ApiHandlerOptions) {
 		this.options = options
-		this.client = new AnthropicVertex({
+		this.clientAnthropic = new AnthropicVertex({
 			projectId: this.options.vertexProjectId,
 			// https://cloud.google.com/vertex-ai/generative-ai/docs/partner-models/use-claude#regions
 			region: this.options.vertexRegion,
 		})
+		this.clientVertex = new VertexAI({
+			project: this.options.vertexProjectId,
+			location: this.options.vertexRegion,
+		})
 	}

 	@withRetry()
@@ -24,40 +30,66 @@ export class VertexHandler implements ApiHandler {
 		const model = this.getModel()
 		const modelId = model.id

-		let budget_tokens = this.options.thinkingBudgetTokens || 0
-		const reasoningOn = budget_tokens !== 0 ? true : false
+		if (modelId.includes("claude")) {
+			let budget_tokens = this.options.thinkingBudgetTokens || 0
+			const reasoningOn = modelId.includes("3-7") && budget_tokens !== 0 ? true : false

-		let stream
-		switch (modelId) {
-			case "claude-3-7-sonnet@20250219":
-			case "claude-3-5-sonnet-v2@20241022":
-			case "claude-3-5-sonnet@20240620":
-			case "claude-3-5-haiku@20241022":
-			case "claude-3-opus@20240229":
-			case "claude-3-haiku@20240307": {
-				// Find indices of user messages for cache control
-				const userMsgIndices = messages.reduce(
-					(acc, msg, index) => (msg.role === "user" ? [...acc, index] : acc),
-					[] as number[],
-				)
-				const lastUserMsgIndex = userMsgIndices[userMsgIndices.length - 1] ?? -1
-				const secondLastMsgUserIndex = userMsgIndices[userMsgIndices.length - 2] ?? -1
+			let stream
+			switch (modelId) {
+				case "claude-3-7-sonnet@20250219":
+				case "claude-3-5-sonnet-v2@20241022":
+				case "claude-3-5-sonnet@20240620":
+				case "claude-3-5-haiku@20241022":
+				case "claude-3-opus@20240229":
+				case "claude-3-haiku@20240307": {
+					// Find indices of user messages for cache control
+					const userMsgIndices = messages.reduce(
+						(acc, msg, index) => (msg.role === "user" ? [...acc, index] : acc),
+						[] as number[],
+					)
+					const lastUserMsgIndex = userMsgIndices[userMsgIndices.length - 1] ?? -1
+					const secondLastMsgUserIndex = userMsgIndices[userMsgIndices.length - 2] ?? -1

-				stream = await this.client.beta.messages.create(
-					{
-						model: modelId,
-						max_tokens: model.info.maxTokens || 8192,
-						thinking: reasoningOn ? { type: "enabled", budget_tokens: budget_tokens } : undefined,
-						temperature: reasoningOn ? undefined : 0,
-						system: [
-							{
-								text: systemPrompt,
-								type: "text",
-								cache_control: { type: "ephemeral" },
-							},
-						],
-						messages: messages.map((message, index) => {
-							if (index === lastUserMsgIndex || index === secondLastMsgUserIndex) {
+					stream = await this.clientAnthropic.beta.messages.create(
+						{
+							model: modelId,
+							max_tokens: model.info.maxTokens || 8192,
+							thinking: reasoningOn ? { type: "enabled", budget_tokens: budget_tokens } : undefined,
+							temperature: reasoningOn ? undefined : 0,
+							system: [
+								{
+									text: systemPrompt,
+									type: "text",
+									cache_control: { type: "ephemeral" },
+								},
+							],
+							messages: messages.map((message, index) => {
+								if (index === lastUserMsgIndex || index === secondLastMsgUserIndex) {
+									return {
+										...message,
+										content:
+											typeof message.content === "string"
+												? [
+														{
+															type: "text",
+															text: message.content,
+															cache_control: {
+																type: "ephemeral",
+															},
+														},
+													]
+												: message.content.map((content, contentIndex) =>
+														contentIndex === message.content.length - 1
+															? {
+																	...content,
+																	cache_control: {
+																		type: "ephemeral",
+																	},
+																}
+															: content,
+													),
+									}
+								}
 								return {
 									...message,
 									content:
@@ -66,121 +98,152 @@ export class VertexHandler implements ApiHandler {
 													{
 														type: "text",
 														text: message.content,
-														cache_control: {
-															type: "ephemeral",
-														},
 													},
 												]
-											: message.content.map((content, contentIndex) =>
-													contentIndex === message.content.length - 1
-														? {
-																...content,
-																cache_control: {
-																	type: "ephemeral",
-																},
-															}
-														: content,
-												),
+											: message.content,
 								}
-							}
-							return {
-								...message,
-								content:
-									typeof message.content === "string"
-										? [
-												{
-													type: "text",
-													text: message.content,
-												},
-											]
-										: message.content,
-							}
-						}),
-						stream: true,
-					},
-					{
-						headers: {},
-					},
-				)
-				break
-			}
-			default: {
-				stream = await this.client.beta.messages.create({
-					model: modelId,
-					max_tokens: model.info.maxTokens || 8192,
-					temperature: 0,
-					system: [
-						{
-							text: systemPrompt,
-							type: "text",
+							}),
+							stream: true,
 						},
-					],
-					messages: messages.map((message) => ({
-						...message,
-						content:
-							typeof message.content === "string"
-								? [
-										{
-											type: "text",
-											text: message.content,
-										},
-									]
-								: message.content,
-					})),
-					stream: true,
-				})
-				break
+						{
+							headers: {},
+						},
+					)
+					break
+				}
+				default: {
+					stream = await this.clientAnthropic.beta.messages.create({
+						model: modelId,
+						max_tokens: model.info.maxTokens || 8192,
+						temperature: 0,
+						system: [
+							{
+								text: systemPrompt,
+								type: "text",
+							},
+						],
+						messages: messages.map((message) => ({
+							...message,
+							content:
+								typeof message.content === "string"
+									? [
+											{
+												type: "text",
+												text: message.content,
+											},
+										]
+									: message.content,
+						})),
+						stream: true,
+					})
+					break
+				}
 			}
-		}
-		for await (const chunk of stream) {
-			switch (chunk.type) {
-				case "message_start":
-					const usage = chunk.message.usage
-					yield {
-						type: "usage",
-						inputTokens: usage.input_tokens || 0,
-						outputTokens: usage.output_tokens || 0,
-						cacheWriteTokens: usage.cache_creation_input_tokens || undefined,
-						cacheReadTokens: usage.cache_read_input_tokens || undefined,
-					}
-					break
-				case "message_delta":
-					yield {
-						type: "usage",
-						inputTokens: 0,
-						outputTokens: chunk.usage.output_tokens || 0,
-					}
-					break
-				case "message_stop":
-					break
-				case "content_block_start":
-					switch (chunk.content_block.type) {
-						case "text":
-							if (chunk.index > 0) {
+			for await (const chunk of stream) {
+				switch (chunk.type) {
+					case "message_start":
+						const usage = chunk.message.usage
+						yield {
+							type: "usage",
+							inputTokens: usage.input_tokens || 0,
+							outputTokens: usage.output_tokens || 0,
+							cacheWriteTokens: usage.cache_creation_input_tokens || undefined,
+							cacheReadTokens: usage.cache_read_input_tokens || undefined,
+						}
+						break
+					case "message_delta":
+						yield {
+							type: "usage",
+							inputTokens: 0,
+							outputTokens: chunk.usage.output_tokens || 0,
+						}
+						break
+					case "message_stop":
+						break
+					case "content_block_start":
+						switch (chunk.content_block.type) {
+							case "text":
+								if (chunk.index > 0) {
+									yield {
+										type: "text",
+										text: "\n",
+									}
+								}
 								yield {
 									type: "text",
-									text: "\n",
+									text: chunk.content_block.text,
 								}
+								break
+						}
+						break
+					case "content_block_delta":
+						switch (chunk.delta.type) {
+							case "text_delta":
+								yield {
+									type: "text",
+									text: chunk.delta.text,
+								}
+								break
+						}
+						break
+					case "content_block_stop":
+						break
+				}
+			}
+		} else {
+			// gemini
+			const generativeModel = this.clientVertex.getGenerativeModel({
+				model: this.getModel().id,
+				systemInstruction: {
+					role: "system",
+					parts: [{ text: systemPrompt }],
+				},
+			})
+			const request = {
+				contents: [
+					{
+						role: "user",
+						parts: messages.map((m) => {
+							if (typeof m.content === "string") {
+								return { text: m.content }
+							} else if (Array.isArray(m.content)) {
+								return {
+									text: m.content
+										.map((block) => {
+											if (typeof block === "string") {
+												return block
+											} else if (block.type === "text") {
+												return block.text
+											} else {
+												console.log("Unsupported block type", block)
+												return ""
+											}
+										})
+										.join(" "),
+								}
+							} else {
+								return { text: "" }
 							}
+						}),
+					},
+				],
+			}
+			const streamingResult = await generativeModel.generateContentStream(request)
+			for await (const chunk of streamingResult.stream) {
+				// If usage data is available, yield it similarly:
+				// yield { type: "usage", inputTokens: 0, outputTokens: 0 }
+				// Otherwise, just yield text:
+				const candidates = chunk.candidates || []
+				for (const candidate of candidates) {
+					for (const part of candidate.content?.parts || []) {
+						if (part.text) {
 							yield {
 								type: "text",
-								text: chunk.content_block.text,
+								text: part.text,
 							}
-							break
+						}
 					}
-					break
-				case "content_block_delta":
-					switch (chunk.delta.type) {
-						case "text_delta":
-							yield {
-								type: "text",
-								text: chunk.delta.text,
-							}
-							break
-					}
-					break
-				case "content_block_stop":
-					break
+				}
 			}
 		}
 	}
@@ -0,0 +1,109 @@
+import { Anthropic } from "@anthropic-ai/sdk"
+import { Message } from "ollama"
+
+export function convertToOllamaMessages(anthropicMessages: Anthropic.Messages.MessageParam[]): Message[] {
+	const ollamaMessages: Message[] = []
+
+	for (const anthropicMessage of anthropicMessages) {
+		if (typeof anthropicMessage.content === "string") {
+			ollamaMessages.push({
+				role: anthropicMessage.role,
+				content: anthropicMessage.content,
+			})
+		} else {
+			if (anthropicMessage.role === "user") {
+				const { nonToolMessages, toolMessages } = anthropicMessage.content.reduce<{
+					nonToolMessages: (Anthropic.TextBlockParam | Anthropic.ImageBlockParam)[]
+					toolMessages: Anthropic.ToolResultBlockParam[]
+				}>(
+					(acc, part) => {
+						if (part.type === "tool_result") {
+							acc.toolMessages.push(part)
+						} else if (part.type === "text" || part.type === "image") {
+							acc.nonToolMessages.push(part)
+						}
+						return acc
+					},
+					{ nonToolMessages: [], toolMessages: [] },
+				)
+
+				// Process tool result messages FIRST since they must follow the tool use messages
+				let toolResultImages: string[] = []
+				toolMessages.forEach((toolMessage) => {
+					// The Anthropic SDK allows tool results to be a string or an array of text and image blocks, enabling rich and structured content. In contrast, the Ollama SDK only supports tool results as a single string, so we map the Anthropic tool result parts into one concatenated string to maintain compatibility.
+					let content: string
+
+					if (typeof toolMessage.content === "string") {
+						content = toolMessage.content
+					} else {
+						content =
+							toolMessage.content
+								?.map((part) => {
+									if (part.type === "image") {
+										toolResultImages.push(`data:${part.source.media_type};base64,${part.source.data}`)
+										return "(see following user message for image)"
+									}
+									return part.text
+								})
+								.join("\n") ?? ""
+					}
+					ollamaMessages.push({
+						role: "user",
+						images: toolResultImages.length > 0 ? toolResultImages : undefined,
+						content: content,
+					})
+				})
+
+				// Process non-tool messages
+				if (nonToolMessages.length > 0) {
+					ollamaMessages.push({
+						role: "user",
+						content: nonToolMessages
+							.map((part) => {
+								if (part.type === "image") {
+									return `data:${part.source.media_type};base64,${part.source.data}`
+								}
+								return part.text
+							})
+							.join("\n"),
+					})
+				}
+			} else if (anthropicMessage.role === "assistant") {
+				const { nonToolMessages, toolMessages } = anthropicMessage.content.reduce<{
+					nonToolMessages: (Anthropic.TextBlockParam | Anthropic.ImageBlockParam)[]
+					toolMessages: Anthropic.ToolUseBlockParam[]
+				}>(
+					(acc, part) => {
+						if (part.type === "tool_use") {
+							acc.toolMessages.push(part)
+						} else if (part.type === "text" || part.type === "image") {
+							acc.nonToolMessages.push(part)
+						} // assistant cannot send tool_result messages
+						return acc
+					},
+					{ nonToolMessages: [], toolMessages: [] },
+				)
+
+				// Process non-tool messages
+				let content: string = ""
+				if (nonToolMessages.length > 0) {
+					content = nonToolMessages
+						.map((part) => {
+							if (part.type === "image") {
+								return "" // impossible as the assistant cannot send images
+							}
+							return part.text
+						})
+						.join("\n")
+				}
+
+				ollamaMessages.push({
+					role: "assistant",
+					content,
+				})
+			}
+		}
+	}
+
+	return ollamaMessages
+}
@@ -4,6 +4,7 @@ import { convertToR1Format } from "./r1-format"
 import { ApiStream, ApiStreamChunk } from "./stream"
 import { Anthropic } from "@anthropic-ai/sdk"
 import OpenAI from "openai"
+import { OpenRouterErrorResponse } from "../providers/types"

 export async function* streamOpenRouterFormatRequest(
 	client: OpenAI,
@@ -146,9 +147,11 @@ export async function* streamOpenRouterFormatRequest(
 	for await (const chunk of stream) {
 		// openrouter returns an error object instead of the openai sdk throwing an error
 		if ("error" in chunk) {
-			const error = chunk.error as { message?: string; code?: number }
+			const error = chunk.error as OpenRouterErrorResponse["error"]
 			console.error(`OpenRouter API Error: ${error?.code} - ${error?.message}`)
-			throw new Error(`OpenRouter API Error ${error?.code}: ${error?.message}`)
+			// Include metadata in the error message if available
+			const metadataStr = error.metadata ? `\nMetadata: ${JSON.stringify(error.metadata, null, 2)}` : ""
+			throw new Error(`OpenRouter API Error ${error.code}: ${error.message}${metadataStr}`)
 		}

 		if (!genId && chunk.id) {
@@ -79,6 +79,7 @@ type GlobalStateKey =
 	| "openAiModelInfo"
 	| "ollamaModelId"
 	| "ollamaBaseUrl"
+	| "ollamaApiOptionsCtxNum"
 	| "lmStudioModelId"
 	| "lmStudioBaseUrl"
 	| "anthropicBaseUrl"
@@ -579,6 +580,7 @@ export class ClineProvider implements vscode.WebviewViewProvider {
 								openAiModelInfo,
 								ollamaModelId,
 								ollamaBaseUrl,
+								ollamaApiOptionsCtxNum,
 								lmStudioModelId,
 								lmStudioBaseUrl,
 								anthropicBaseUrl,
@@ -624,6 +626,7 @@ export class ClineProvider implements vscode.WebviewViewProvider {
 							await this.updateGlobalState("openAiModelInfo", openAiModelInfo)
 							await this.updateGlobalState("ollamaModelId", ollamaModelId)
 							await this.updateGlobalState("ollamaBaseUrl", ollamaBaseUrl)
+							await this.updateGlobalState("ollamaApiOptionsCtxNum", ollamaApiOptionsCtxNum)
 							await this.updateGlobalState("lmStudioModelId", lmStudioModelId)
 							await this.updateGlobalState("lmStudioBaseUrl", lmStudioBaseUrl)
 							await this.updateGlobalState("anthropicBaseUrl", anthropicBaseUrl)
@@ -1901,6 +1904,7 @@ Here is the project's README to help you get started:\n\n${mcpDetails.readmeCont
 			openAiModelInfo,
 			ollamaModelId,
 			ollamaBaseUrl,
+			ollamaApiOptionsCtxNum,
 			lmStudioModelId,
 			lmStudioBaseUrl,
 			anthropicBaseUrl,
@@ -1959,6 +1963,7 @@ Here is the project's README to help you get started:\n\n${mcpDetails.readmeCont
 			this.getGlobalState("openAiModelInfo") as Promise<ModelInfo | undefined>,
 			this.getGlobalState("ollamaModelId") as Promise<string | undefined>,
 			this.getGlobalState("ollamaBaseUrl") as Promise<string | undefined>,
+			this.getGlobalState("ollamaApiOptionsCtxNum") as Promise<string | undefined>,
 			this.getGlobalState("lmStudioModelId") as Promise<string | undefined>,
 			this.getGlobalState("lmStudioBaseUrl") as Promise<string | undefined>,
 			this.getGlobalState("anthropicBaseUrl") as Promise<string | undefined>,
@@ -2040,6 +2045,7 @@ Here is the project's README to help you get started:\n\n${mcpDetails.readmeCont
 				openAiModelInfo,
 				ollamaModelId,
 				ollamaBaseUrl,
+				ollamaApiOptionsCtxNum,
 				lmStudioModelId,
 				lmStudioBaseUrl,
 				anthropicBaseUrl,
@@ -46,6 +46,7 @@ export interface ApiHandlerOptions {
 	openAiModelInfo?: ModelInfo
 	ollamaModelId?: string
 	ollamaBaseUrl?: string
+	ollamaApiOptionsCtxNum?: string
 	lmStudioModelId?: string
 	lmStudioBaseUrl?: string
 	geminiApiKey?: string
@@ -298,6 +299,78 @@ export const vertexModels = {
 		cacheWritesPrice: 0.3,
 		cacheReadsPrice: 0.03,
 	},
+	"gemini-2.0-flash-001": {
+		maxTokens: 8192,
+		contextWindow: 1_048_576,
+		supportsImages: true,
+		supportsPromptCache: false,
+		inputPrice: 0.1,
+		outputPrice: 0.4,
+	},
+	"gemini-2.0-flash-thinking-exp-1219": {
+		maxTokens: 8192,
+		contextWindow: 32_767,
+		supportsImages: true,
+		supportsPromptCache: false,
+		inputPrice: 0,
+		outputPrice: 0,
+	},
+	"gemini-2.0-flash-exp": {
+		maxTokens: 8192,
+		contextWindow: 1_048_576,
+		supportsImages: true,
+		supportsPromptCache: false,
+		inputPrice: 0,
+		outputPrice: 0,
+	},
+	"gemini-exp-1206": {
+		maxTokens: 8192,
+		contextWindow: 2_097_152,
+		supportsImages: true,
+		supportsPromptCache: false,
+		inputPrice: 0,
+		outputPrice: 0,
+	},
+	"gemini-1.5-flash-002": {
+		maxTokens: 8192,
+		contextWindow: 1_048_576,
+		supportsImages: true,
+		supportsPromptCache: false,
+		inputPrice: 0,
+		outputPrice: 0,
+	},
+	"gemini-1.5-flash-exp-0827": {
+		maxTokens: 8192,
+		contextWindow: 1_048_576,
+		supportsImages: true,
+		supportsPromptCache: false,
+		inputPrice: 0,
+		outputPrice: 0,
+	},
+	"gemini-1.5-flash-8b-exp-0827": {
+		maxTokens: 8192,
+		contextWindow: 1_048_576,
+		supportsImages: true,
+		supportsPromptCache: false,
+		inputPrice: 0,
+		outputPrice: 0,
+	},
+	"gemini-1.5-pro-002": {
+		maxTokens: 8192,
+		contextWindow: 2_097_152,
+		supportsImages: true,
+		supportsPromptCache: false,
+		inputPrice: 0,
+		outputPrice: 0,
+	},
+	"gemini-1.5-pro-exp-0827": {
+		maxTokens: 8192,
+		contextWindow: 2_097_152,
+		supportsImages: true,
+		supportsPromptCache: false,
+		inputPrice: 0,
+		outputPrice: 0,
+	},
 } as const satisfies Record<string, ModelInfo>

 export const openAiModelInfoSaneDefaults: ModelInfo = {
@@ -815,7 +815,7 @@ export const ChatRowContent = ({ message, isExpanded, onToggleExpand, lastModifi
 									{isExpanded ? (
 										<div style={{ marginTop: -3 }}>
 											<span style={{ fontWeight: "bold", display: "block", marginBottom: "4px" }}>
-												Reasoning
+												Thinking
 												<span
 													className="codicon codicon-chevron-down"
 													style={{
@@ -1173,6 +1173,13 @@ const ApiOptions = ({ showModelOptions, apiErrorMessage, modelIdErrorMessage, is
 						placeholder={"e.g. llama3.1"}>
 						<span style={{ fontWeight: 500 }}>Model ID</span>
 					</VSCodeTextField>
+					<VSCodeTextField
+						value={apiConfiguration?.ollamaApiOptionsCtxNum || "32768"}
+						style={{ width: "100%" }}
+						onInput={handleInputChange("ollamaApiOptionsCtxNum")}
+						placeholder={"e.g. 32768"}>
+						<span style={{ fontWeight: 500 }}>Model Context Window</span>
+					</VSCodeTextField>
 					{ollamaModels.length > 0 && (
 						<VSCodeRadioGroup
 							value={
@@ -59,7 +59,7 @@ const RangeInput = styled.input<{ $value: number; $min: number; $max: number }>`
 		border-radius: 50%;
 		background: var(--vscode-foreground);
 		cursor: pointer;
-		border: 2px solid var(--vscode-progressBar-background);
+		border: 0px solid var(--vscode-progressBar-background);
 		box-shadow: 0 2px 4px rgba(0, 0, 0, 0.2);
 	}

@@ -180,7 +180,9 @@ export function getContextMenuOptions(
 		const seen = new Set()
 		const deduped = allItems.filter((item) => {
 			const key = `${item.type}-${item.value}`
-			if (seen.has(key)) return false
+			if (seen.has(key)) {
+				return false
+			}
 			seen.add(key)
 			return true
 		})
@@ -195,18 +197,26 @@ export function shouldShowContextMenu(text: string, position: number): boolean {
 	const beforeCursor = text.slice(0, position)
 	const atIndex = beforeCursor.lastIndexOf("@")

-	if (atIndex === -1) return false
+	if (atIndex === -1) {
+		return false
+	}

 	const textAfterAt = beforeCursor.slice(atIndex + 1)

 	// Check if there's any whitespace after the '@'
-	if (/\s/.test(textAfterAt)) return false
+	if (/\s/.test(textAfterAt)) {
+		return false
+	}

 	// Don't show the menu if it's a URL
-	if (textAfterAt.toLowerCase().startsWith("http")) return false
+	if (textAfterAt.toLowerCase().startsWith("http")) {
+		return false
+	}

 	// Don't show the menu if it's a problems or terminal
-	if (textAfterAt.toLowerCase().startsWith("problems") || textAfterAt.toLowerCase().startsWith("terminal")) return false
+	if (textAfterAt.toLowerCase().startsWith("problems") || textAfterAt.toLowerCase().startsWith("terminal")) {
+		return false
+	}

 	// NOTE: it's okay that menu shows when there's trailing punctuation since user could be inputting a path with marks

@@ -50,8 +50,10 @@ export const useShortcut = (shortcut: string, callback: any, options = { disable

 				if (Object.keys(modifierMap).includes(keyArray[0])) {
 					const finalKey = keyArray.pop()
+					if (!finalKey) return

-					if (keyArray.every((k) => modifierMap[k]) && finalKey === event.key) {
+					if (keyArray.every((k) => modifierMap[k]) && finalKey.toLowerCase() === event.key.toLowerCase()) {
+						event.preventDefault()
 						return callbackRef.current(event)
 					}
 				} else {
@@ -35,7 +35,9 @@ export function findMatchingResourceOrTemplate(
 ): McpResource | McpResourceTemplate | undefined {
 	// First try to find an exact resource match
 	const exactMatch = resources.find((resource) => resource.uri === uri)
-	if (exactMatch) return exactMatch
+	if (exactMatch) {
+		return exactMatch
+	}

 	// If no exact match, try to find a matching template
 	return findMatchingTemplate(uri, templates)
@@ -36,7 +36,5 @@ export function useDebounceEffect(effect: VoidFn, delay: number, deps: any[]) {

 		// We want to re‐schedule if any item in `deps` changed,
 		// or if `delay` changed.
-
-		// eslint-disable-next-line react-hooks/exhaustive-deps
 	}, [delay, ...deps])
 }
@@ -35,7 +35,9 @@ export function hexToRGB(hexColor: string): { r: number; g: number; b: number }

 export function colorToHex(colorVar: string): string {
 	const value = getComputedStyle(document.documentElement).getPropertyValue(colorVar).trim()
-	if (value.startsWith("#")) return value.slice(0, 7)
+	if (value.startsWith("#")) {
+		return value.slice(0, 7)
+	}

 	const rgbValues = value.match(/\d+/g)?.slice(0, 3).map(Number) || []
 	return `#${rgbValues.map((x) => x.toString(16).padStart(2, "0")).join("")}`