Feat/nvidia nim provider support enhance + more...

.husky/pre-commit

@@ -1,27 +1,12 @@
+#!/usr/bin/env sh
 branch="$(git rev-parse --abbrev-ref HEAD)"
 
 if [ "$branch" = "main" ]; then
   echo "You can't commit directly to main - please check out a branch."
   exit 1
 fi
 
-# Detect if running on Windows and use pnpm.cmd, otherwise use pnpm.
-if [ "$OS" = "Windows_NT" ]; then
-  pnpm_cmd="pnpm.cmd"
-else
-  if command -v pnpm >/dev/null 2>&1; then
-    pnpm_cmd="pnpm"
-  else
-    pnpm_cmd="npx pnpm"
-  fi
-fi
-
-# Detect if running on Windows and use npx.cmd, otherwise use npx.
-if [ "$OS" = "Windows_NT" ]; then
-  npx_cmd="npx.cmd"
-else
-  npx_cmd="npx"
-fi
-
-$npx_cmd lint-staged
-$pnpm_cmd lint
+# Use npx to run lint-staged and pnpm for lint
+# Git Bash on Windows automatically resolves pnpm to pnpm.cmd
+npx lint-staged
+pnpm lint

.husky/pre-push

@@ -1,33 +1,24 @@
+#!/usr/bin/env sh
 branch="$(git rev-parse --abbrev-ref HEAD)"
 
 if [ "$branch" = "main" ]; then
   echo "You can't push directly to main - please check out a branch."
   exit 1
 fi
 
-# Detect if running on Windows and use pnpm.cmd, otherwise use pnpm.
-if [ "$OS" = "Windows_NT" ]; then
-  pnpm_cmd="pnpm.cmd"
-else
-  if command -v pnpm >/dev/null 2>&1; then
-    pnpm_cmd="pnpm"
-  else
-    pnpm_cmd="npx pnpm"
-  fi
-fi
-
-$pnpm_cmd run check-types
+# Git Bash on Windows automatically resolves pnpm to pnpm.cmd
+pnpm run check-types
 
 # Use dotenvx to securely load .env.local and run commands that depend on it
 if [ -f ".env.local" ]; then
   # Check if RUN_TESTS_ON_PUSH is set to true and run tests with dotenvx
   if npx dotenvx get RUN_TESTS_ON_PUSH -f .env.local 2>/dev/null | grep -q "^true$"; then
-    npx dotenvx run -f .env.local -- $pnpm_cmd run test
+    npx dotenvx run -f .env.local -- pnpm run test
   fi
 else
   # Fallback: run tests if RUN_TESTS_ON_PUSH is set in regular environment
   if [ "$RUN_TESTS_ON_PUSH" = "true" ]; then
-    $pnpm_cmd run test
+    pnpm run test
   fi
 fi
 

apps/cli/turbo.json

@@ -0,0 +1,9 @@
+{
+	"$schema": "https://turbo.build/schema.json",
+	"extends": ["//"],
+	"tasks": {
+		"build": {
+			"dependsOn": ["@roo-code/types#build", "@roo-code/core#build", "@roo-code/vscode-shim#build"]
+		}
+	}
+}

apps/web-evals/next-env.d.ts

@@ -1,6 +1,6 @@
 /// <reference types="next" />
 /// <reference types="next/image-types/global" />
-import "./.next/dev/types/routes.d.ts"
+import "./.next/types/routes.d.ts"
 
 // NOTE: This file should not be edited
 // see https://nextjs.org/docs/app/api-reference/config/typescript for more information.

apps/web-evals/next.config.ts

@@ -1,7 +1,11 @@
+import path from "path"
 import type { NextConfig } from "next"
 
 const nextConfig: NextConfig = {
-	turbopack: {},
+	turbopack: {
+		root: path.join(__dirname, "../.."),
+	},
+	transpilePackages: ["@roo-code/types"],
 }
 
 export default nextConfig

apps/web-evals/package.json

@@ -54,6 +54,7 @@
 		"@roo-code/config-typescript": "workspace:^",
 		"@tailwindcss/postcss": "^4",
 		"@types/archiver": "^7.0.0",
+		"@types/node": "^25.5.2",
 		"@types/ps-tree": "^1.1.6",
 		"@types/react": "^18.3.23",
 		"@types/react-dom": "^18.3.5",

package.json

@@ -1,6 +1,6 @@
 {
 	"name": "roo-code",
-	"packageManager": "pnpm@10.8.1",
+	"packageManager": "pnpm@10.33.0",
 	"engines": {
 		"node": "20.19.2"
 	},

packages/core/package.json

@@ -16,7 +16,7 @@
 	},
 	"dependencies": {
 		"@roo-code/types": "workspace:^",
-		"esbuild": "^0.25.0",
+		"esbuild-wasm": "^0.25.12",
 		"execa": "^9.5.2",
 		"ignore": "^7.0.3",
 		"openai": "^5.12.2",

packages/types/package.json

@@ -5,8 +5,8 @@
 	"main": "./dist/index.cjs",
 	"exports": {
 		".": {
-			"types": "./src/index.ts",
-			"import": "./src/index.ts",
+			"types": "./dist/index.d.ts",
+			"import": "./dist/index.js",
 			"require": {
 				"types": "./dist/index.d.cts",
 				"default": "./dist/index.cjs"

packages/types/src/mcp.ts

@@ -99,22 +99,35 @@ export type McpResourceResponse = {
 	}>
 }
 
+// Annotations for content items in tool call responses
+export type McpContentAnnotations = {
+	audience?: Array<"user" | "assistant">
+	priority?: number
+	lastModified?: string
+}
+
 export type McpToolCallResponse = {
 	_meta?: Record<string, any> // eslint-disable-line @typescript-eslint/no-explicit-any
 	content: Array<
 		| {
 				type: "text"
 				text: string
+				annotations?: McpContentAnnotations
+				_meta?: Record<string, any> // eslint-disable-line @typescript-eslint/no-explicit-any
 		  }
 		| {
 				type: "image"
 				data: string
 				mimeType: string
+				annotations?: McpContentAnnotations
+				_meta?: Record<string, any> // eslint-disable-line @typescript-eslint/no-explicit-any
 		  }
 		| {
 				type: "audio"
 				data: string
 				mimeType: string
+				annotations?: McpContentAnnotations
+				_meta?: Record<string, any> // eslint-disable-line @typescript-eslint/no-explicit-any
 		  }
 		| {
 				type: "resource"
@@ -124,6 +137,25 @@ export type McpToolCallResponse = {
 					text?: string
 					blob?: string
 				}
+				annotations?: McpContentAnnotations
+				_meta?: Record<string, any> // eslint-disable-line @typescript-eslint/no-explicit-any
+		  }
+		| {
+				type: "resource_link"
+				uri: string
+				name: string
+				description?: string
+				mimeType?: string
+				size?: number
+				annotations?: McpContentAnnotations
+				_meta?: Record<string, any> // eslint-disable-line @typescript-eslint/no-explicit-any
+				icons?: Array<{
+					src: string
+					mimeType?: string
+					sizes?: string[]
+					theme?: "light" | "dark"
+				}>
+				title?: string
 		  }
 	>
 	isError?: boolean

packages/types/src/provider-settings.ts

@@ -15,6 +15,7 @@ import {
 	openAiNativeModels,
 	qwenCodeModels,
 	sambaNovaModels,
+	nvidiaModels,
 	vertexModels,
 	vscodeLlmModels,
 	xaiModels,
@@ -117,6 +118,7 @@ export const providerNames = [
 	"qwen-code",
 	"roo",
 	"sambanova",
+	"nvidia",
 	"vertex",
 	"xai",
 	"zai",
@@ -350,6 +352,11 @@ const sambaNovaSchema = apiModelIdProviderModelSchema.extend({
 	sambaNovaApiKey: z.string().optional(),
 })
 
+const nvidiaSchema = apiModelIdProviderModelSchema.extend({
+	nvidiaApiKey: z.string().optional(),
+	nvidiaBaseUrl: z.string().optional(),
+})
+
 export const zaiApiLineSchema = z.enum(["international_coding", "china_coding", "international_api", "china_api"])
 
 export type ZaiApiLine = z.infer<typeof zaiApiLineSchema>
@@ -409,6 +416,7 @@ export const providerSettingsSchemaDiscriminated = z.discriminatedUnion("apiProv
 	basetenSchema.merge(z.object({ apiProvider: z.literal("baseten") })),
 	litellmSchema.merge(z.object({ apiProvider: z.literal("litellm") })),
 	sambaNovaSchema.merge(z.object({ apiProvider: z.literal("sambanova") })),
+	nvidiaSchema.merge(z.object({ apiProvider: z.literal("nvidia") })),
 	zaiSchema.merge(z.object({ apiProvider: z.literal("zai") })),
 	fireworksSchema.merge(z.object({ apiProvider: z.literal("fireworks") })),
 	qwenCodeSchema.merge(z.object({ apiProvider: z.literal("qwen-code") })),
@@ -442,6 +450,7 @@ export const providerSettingsSchema = z.object({
 	...basetenSchema.shape,
 	...litellmSchema.shape,
 	...sambaNovaSchema.shape,
+	...nvidiaSchema.shape,
 	...zaiSchema.shape,
 	...fireworksSchema.shape,
 	...qwenCodeSchema.shape,
@@ -517,6 +526,7 @@ export const modelIdKeysByProvider: Record<TypicalProvider, ModelIdKey> = {
 	baseten: "apiModelId",
 	litellm: "litellmModelId",
 	sambanova: "apiModelId",
+	nvidia: "apiModelId",
 	zai: "apiModelId",
 	fireworks: "apiModelId",
 	roo: "apiModelId",
@@ -617,6 +627,11 @@ export const MODELS_BY_PROVIDER: Record<
 		label: "SambaNova",
 		models: Object.keys(sambaNovaModels),
 	},
+	nvidia: {
+		id: "nvidia",
+		label: "NVIDIA NIM",
+		models: Object.keys(nvidiaModels),
+	},
 	vertex: {
 		id: "vertex",
 		label: "GCP Vertex AI",

packages/types/src/providers/index.ts

@@ -17,6 +17,7 @@ export * from "./qwen-code.js"
 export * from "./requesty.js"
 export * from "./roo.js"
 export * from "./sambanova.js"
+export * from "./nvidia.js"
 export * from "./unbound.js"
 export * from "./vertex.js"
 export * from "./vscode-llm.js"
@@ -40,6 +41,7 @@ import { qwenCodeDefaultModelId } from "./qwen-code.js"
 import { requestyDefaultModelId } from "./requesty.js"
 import { rooDefaultModelId } from "./roo.js"
 import { sambaNovaDefaultModelId } from "./sambanova.js"
+import { nvidiaDefaultModelId } from "./nvidia.js"
 import { unboundDefaultModelId } from "./unbound.js"
 import { vertexDefaultModelId } from "./vertex.js"
 import { vscodeLlmDefaultModelId } from "./vscode-llm.js"
@@ -101,6 +103,8 @@ export function getProviderDefaultModelId(
 			return vscodeLlmDefaultModelId
 		case "sambanova":
 			return sambaNovaDefaultModelId
+		case "nvidia":
+			return nvidiaDefaultModelId
 		case "fireworks":
 			return fireworksDefaultModelId
 		case "roo":

packages/types/src/providers/nvidia.ts

@@ -0,0 +1,93 @@
+import type { ModelInfo } from "../model.js"
+
+// NVIDIA NIM API model IDs
+// Reference: https://build.nvidia.com/explore/discover/models
+export type NvidiaModelId =
+	| "deepseek-ai/deepseek-r1"
+	| "google/gemma-4-31b-it"
+	| "meta/llama-3.1-405b-instruct"
+	| "meta/llama-3.3-70b-instruct"
+	| "z-ai/glm5"
+	| "qwen/qwen3.5-397b-a17b"
+	| "qwen/qwen3.5-122b-a10b"
+
+export const nvidiaDefaultModelId: NvidiaModelId = "deepseek-ai/deepseek-r1"
+
+export const nvidiaModels = {
+	// DeepSeek R1 - reasoning model with enable_thinking support
+	"deepseek-ai/deepseek-r1": {
+		maxTokens: 16384,
+		contextWindow: 128000,
+		supportsImages: false,
+		supportsPromptCache: false,
+		// Critical: This flag enables the reasoning toggle in UI
+		supportsReasoningBinary: true,
+		inputPrice: 0.0, // NVIDIA NIM pricing varies by deployment
+		outputPrice: 0.0,
+		description:
+			"DeepSeek R1 reasoning model via NVIDIA NIM API. Supports chain-of-thought reasoning with enable_thinking parameter.",
+	},
+	// Google Gemma 4 - reasoning model with enable_thinking support
+	"google/gemma-4-31b-it": {
+		maxTokens: 16384,
+		contextWindow: 128000,
+		supportsImages: false,
+		supportsPromptCache: false,
+		supportsReasoningBinary: true,
+		inputPrice: 0.0,
+		outputPrice: 0.0,
+		description:
+			"Google Gemma 4 31B reasoning model via NVIDIA NIM API. Supports chain-of-thought reasoning with enable_thinking parameter.",
+	},
+	// Llama models - no reasoning support, included for completeness
+	"meta/llama-3.1-405b-instruct": {
+		maxTokens: 8192,
+		contextWindow: 128000,
+		supportsImages: true,
+		supportsPromptCache: false,
+		inputPrice: 0.0,
+		outputPrice: 0.0,
+		description: "Meta Llama 3.1 405B Instruct model via NVIDIA NIM API.",
+	},
+	"meta/llama-3.3-70b-instruct": {
+		maxTokens: 8192,
+		contextWindow: 128000,
+		supportsImages: true,
+		supportsPromptCache: false,
+		inputPrice: 0.0,
+		outputPrice: 0.0,
+		description: "Meta Llama 3.3 70B Instruct model via NVIDIA NIM API.",
+	},
+	// GLM5 - Z.AI reasoning model with enable_thinking support
+	"z-ai/glm5": {
+		maxTokens: 32768,
+		contextWindow: 204800,
+		supportsImages: false,
+		supportsPromptCache: false,
+		supportsReasoningBinary: true,
+		inputPrice: 1.0,
+		outputPrice: 3.2,
+		description: "Z.AI GLM5 reasoning model via NVIDIA NIM API. Supports chain-of-thought reasoning.",
+	},
+	// Qwen 3.5 models - reasoning models with enable_thinking support
+	"qwen/qwen3.5-397b-a17b": {
+		maxTokens: 32768,
+		contextWindow: 262144,
+		supportsImages: false,
+		supportsPromptCache: false,
+		supportsReasoningBinary: true,
+		inputPrice: 0.0,
+		outputPrice: 0.0,
+		description: "Qwen 3.5 397B reasoning model via NVIDIA NIM API. Supports chain-of-thought reasoning.",
+	},
+	"qwen/qwen3.5-122b-a10b": {
+		maxTokens: 32768,
+		contextWindow: 262144,
+		supportsImages: false,
+		supportsPromptCache: false,
+		supportsReasoningBinary: true,
+		inputPrice: 0.0,
+		outputPrice: 0.0,
+		description: "Qwen 3.5 122B reasoning model via NVIDIA NIM API. Supports chain-of-thought reasoning.",
+	},
+} as const satisfies Record<string, ModelInfo>