nvidia-nim-mcp 1.0.0

package/dist/logger.js

@@ -0,0 +1,49 @@
+import winston from "winston";
+const { combine, timestamp, errors, json, colorize, simple } = winston.format;
+// Initialize logger with default configuration
+export let logger = winston.createLogger({
+    level: "info",
+    format: combine(timestamp({ format: "YYYY-MM-DD HH:mm:ss.SSS" }), errors({ stack: true }), combine(colorize({ all: true }), simple())),
+    defaultMeta: {
+        service: "nvidia-nim-mcp",
+        version: "1.0.0",
+    },
+    transports: [
+        new winston.transports.Console({
+            stderrLevels: ["error", "warn"],
+        }),
+    ],
+});
+export function initLogger(options) {
+    const { logLevel, isProduction, mcpServerVersion } = options;
+    logger = winston.createLogger({
+        level: logLevel,
+        format: combine(timestamp({ format: "YYYY-MM-DD HH:mm:ss.SSS" }), errors({ stack: true }), isProduction
+            ? json()
+            : combine(colorize({ all: true }), simple())),
+        defaultMeta: {
+            service: "nvidia-nim-mcp",
+            version: mcpServerVersion,
+        },
+        transports: [
+            new winston.transports.Console({
+                stderrLevels: ["error", "warn"],
+            }),
+        ],
+    });
+    // Log to file in production
+    if (isProduction) {
+        logger.add(new winston.transports.File({
+            filename: "logs/error.log",
+            level: "error",
+            maxsize: 10 * 1024 * 1024, // 10MB
+            maxFiles: 5,
+        }));
+        logger.add(new winston.transports.File({
+            filename: "logs/combined.log",
+            maxsize: 10 * 1024 * 1024, // 10MB
+            maxFiles: 10,
+        }));
+    }
+    return logger;
+}

package/dist/models.d.ts

@@ -0,0 +1,14 @@
+export interface NIMModel {
+    id: string;
+    name: string;
+    description: string;
+    category: ModelCategory;
+    contextLength: number;
+    supportsStreaming: boolean;
+    supportsFunctionCalling: boolean;
+    supportsVision?: boolean;
+}
+export type ModelCategory = "language" | "embedding" | "reranking" | "vision" | "code" | "multimodal";
+export declare const NIM_MODELS: Record<string, NIMModel>;
+export declare function getModelsByCategory(category: ModelCategory): NIMModel[];
+export declare function getModel(modelId: string): NIMModel | undefined;

package/dist/models.js

@@ -0,0 +1,202 @@
+export const NIM_MODELS = {
+    // === LLMs ===
+    "meta/llama-3.1-405b-instruct": {
+        id: "meta/llama-3.1-405b-instruct",
+        name: "Llama 3.1 405B Instruct",
+        description: "Meta's largest Llama 3.1 model, best for complex reasoning",
+        category: "language",
+        contextLength: 128000,
+        supportsStreaming: true,
+        supportsFunctionCalling: true,
+    },
+    "meta/llama-3.1-70b-instruct": {
+        id: "meta/llama-3.1-70b-instruct",
+        name: "Llama 3.1 70B Instruct",
+        description: "Balanced performance and efficiency",
+        category: "language",
+        contextLength: 128000,
+        supportsStreaming: true,
+        supportsFunctionCalling: true,
+    },
+    "meta/llama-3.1-8b-instruct": {
+        id: "meta/llama-3.1-8b-instruct",
+        name: "Llama 3.1 8B Instruct",
+        description: "Fast and efficient for simpler tasks",
+        category: "language",
+        contextLength: 128000,
+        supportsStreaming: true,
+        supportsFunctionCalling: true,
+    },
+    "meta/llama-3.2-90b-vision-instruct": {
+        id: "meta/llama-3.2-90b-vision-instruct",
+        name: "Llama 3.2 90B Vision Instruct",
+        description: "Multimodal model with vision capabilities",
+        category: "vision",
+        contextLength: 128000,
+        supportsStreaming: true,
+        supportsFunctionCalling: false,
+        supportsVision: true,
+    },
+    "meta/llama-3.2-11b-vision-instruct": {
+        id: "meta/llama-3.2-11b-vision-instruct",
+        name: "Llama 3.2 11B Vision Instruct",
+        description: "Efficient multimodal model",
+        category: "vision",
+        contextLength: 128000,
+        supportsStreaming: true,
+        supportsFunctionCalling: false,
+        supportsVision: true,
+    },
+    "mistralai/mistral-large-2-instruct": {
+        id: "mistralai/mistral-large-2-instruct",
+        name: "Mistral Large 2",
+        description: "Mistral's flagship model with excellent reasoning",
+        category: "language",
+        contextLength: 131072,
+        supportsStreaming: true,
+        supportsFunctionCalling: true,
+    },
+    "mistralai/mixtral-8x22b-instruct-v0.1": {
+        id: "mistralai/mixtral-8x22b-instruct-v0.1",
+        name: "Mixtral 8x22B Instruct",
+        description: "MoE model with excellent performance",
+        category: "language",
+        contextLength: 65536,
+        supportsStreaming: true,
+        supportsFunctionCalling: true,
+    },
+    "mistralai/mixtral-8x7b-instruct-v0.1": {
+        id: "mistralai/mixtral-8x7b-instruct-v0.1",
+        name: "Mixtral 8x7B Instruct",
+        description: "Efficient MoE model",
+        category: "language",
+        contextLength: 32768,
+        supportsStreaming: true,
+        supportsFunctionCalling: false,
+    },
+    "microsoft/phi-3.5-mini-instruct": {
+        id: "microsoft/phi-3.5-mini-instruct",
+        name: "Phi-3.5 Mini Instruct",
+        description: "Microsoft's efficient small model",
+        category: "language",
+        contextLength: 128000,
+        supportsStreaming: true,
+        supportsFunctionCalling: false,
+    },
+    "microsoft/phi-3-medium-128k-instruct": {
+        id: "microsoft/phi-3-medium-128k-instruct",
+        name: "Phi-3 Medium 128K Instruct",
+        description: "Microsoft's medium model with long context",
+        category: "language",
+        contextLength: 128000,
+        supportsStreaming: true,
+        supportsFunctionCalling: false,
+    },
+    "google/gemma-2-27b-it": {
+        id: "google/gemma-2-27b-it",
+        name: "Gemma 2 27B IT",
+        description: "Google's Gemma 2 model",
+        category: "language",
+        contextLength: 8192,
+        supportsStreaming: true,
+        supportsFunctionCalling: false,
+    },
+    "google/gemma-2-9b-it": {
+        id: "google/gemma-2-9b-it",
+        name: "Gemma 2 9B IT",
+        description: "Google's efficient Gemma 2 model",
+        category: "language",
+        contextLength: 8192,
+        supportsStreaming: true,
+        supportsFunctionCalling: false,
+    },
+    "qwen/qwen2.5-72b-instruct": {
+        id: "qwen/qwen2.5-72b-instruct",
+        name: "Qwen 2.5 72B Instruct",
+        description: "Alibaba's Qwen 2.5 large model",
+        category: "language",
+        contextLength: 131072,
+        supportsStreaming: true,
+        supportsFunctionCalling: true,
+    },
+    "qwen/qwen2.5-coder-32b-instruct": {
+        id: "qwen/qwen2.5-coder-32b-instruct",
+        name: "Qwen 2.5 Coder 32B",
+        description: "Specialized coding model",
+        category: "code",
+        contextLength: 131072,
+        supportsStreaming: true,
+        supportsFunctionCalling: false,
+    },
+    "nvidia/llama-3.1-nemotron-70b-instruct": {
+        id: "nvidia/llama-3.1-nemotron-70b-instruct",
+        name: "Nemotron 70B Instruct",
+        description: "NVIDIA's fine-tuned Llama with RLHF",
+        category: "language",
+        contextLength: 131072,
+        supportsStreaming: true,
+        supportsFunctionCalling: false,
+    },
+    "z-ai/glm-4-9b-chat": {
+        id: "z-ai/glm-4-9b-chat",
+        name: "GLM-4 9B Chat",
+        description: "Zhipu AI's GLM-4 model, strong multilingual and reasoning capabilities",
+        category: "language",
+        contextLength: 128000,
+        supportsStreaming: true,
+        supportsFunctionCalling: false,
+    },
+    "z-ai/glm5": {
+        id: "z-ai/glm5",
+        name: "GLM-5",
+        description: "Zhipu AI's GLM-5 model specialized in software development and architecture - excels at code generation, debugging, refactoring, and system design",
+        category: "code",
+        contextLength: 128000,
+        supportsStreaming: true,
+        supportsFunctionCalling: true,
+    },
+    // === Embeddings ===
+    "nvidia/nv-embedqa-e5-v5": {
+        id: "nvidia/nv-embedqa-e5-v5",
+        name: "NV-EmbedQA E5 v5",
+        description: "NVIDIA embedding model optimized for Q&A retrieval",
+        category: "embedding",
+        contextLength: 512,
+        supportsStreaming: false,
+        supportsFunctionCalling: false,
+    },
+    "nvidia/nv-embed-v1": {
+        id: "nvidia/nv-embed-v1",
+        name: "NV-Embed v1",
+        description: "NVIDIA general purpose embedding model",
+        category: "embedding",
+        contextLength: 4096,
+        supportsStreaming: false,
+        supportsFunctionCalling: false,
+    },
+    "baai/bge-m3": {
+        id: "baai/bge-m3",
+        name: "BGE-M3",
+        description: "Multilingual embedding model",
+        category: "embedding",
+        contextLength: 8192,
+        supportsStreaming: false,
+        supportsFunctionCalling: false,
+    },
+    // === Reranking ===
+    "nvidia/nv-rerankqa-mistral-4b-v3": {
+        id: "nvidia/nv-rerankqa-mistral-4b-v3",
+        name: "NV-RerankQA Mistral 4B v3",
+        description: "NVIDIA reranking model for RAG pipelines",
+        category: "reranking",
+        contextLength: 4096,
+        supportsStreaming: false,
+        supportsFunctionCalling: false,
+    },
+};
+export function getModelsByCategory(category) {
+    return Object.values(NIM_MODELS).filter((m) => m.category === category);
+}
+export function getModel(modelId) {
+    return NIM_MODELS[modelId];
+}