@juspay/neurolink 7.33.2 → 7.33.4

package/dist/lib/neurolink.js

@@ -16,6 +16,7 @@ catch {
 import { AIProviderFactory } from "./core/factory.js";
 import { mcpLogger } from "./utils/logger.js";
 import { SYSTEM_LIMITS } from "./core/constants.js";
+import { NANOSECOND_TO_MS_DIVISOR, MCP_TIMEOUTS, SERVER_CONFIG, TOOL_TIMEOUTS, RETRY_ATTEMPTS, RETRY_DELAYS, CIRCUIT_BREAKER, CIRCUIT_BREAKER_RESET_MS, MEMORY_THRESHOLDS, PROVIDER_TIMEOUTS, PERFORMANCE_THRESHOLDS, } from "./constants/index.js";
 import pLimit from "p-limit";
 import { toolRegistry } from "./mcp/toolRegistry.js";
 import { logger } from "./utils/logger.js";
@@ -164,7 +165,7 @@ export class NeuroLink {
                 elapsedMs: Date.now() - constructorStartTime,
                 elapsedNs: (process.hrtime.bigint() - constructorHrTimeStart).toString(),
                 registrySetupDurationNs: registrySetupDurationNs.toString(),
-                registrySetupDurationMs: Number(registrySetupDurationNs) / 1000000,
+                registrySetupDurationMs: Number(registrySetupDurationNs) / NANOSECOND_TO_MS_DIVISOR,
                 enableManualMCP: false,
                 message: "ProviderRegistry configured successfully with security settings",
             });
@@ -179,7 +180,7 @@ export class NeuroLink {
                 elapsedMs: Date.now() - constructorStartTime,
                 elapsedNs: (process.hrtime.bigint() - constructorHrTimeStart).toString(),
                 registrySetupDurationNs: registrySetupDurationNs.toString(),
-                registrySetupDurationMs: Number(registrySetupDurationNs) / 1000000,
+                registrySetupDurationMs: Number(registrySetupDurationNs) / NANOSECOND_TO_MS_DIVISOR,
                 error: error instanceof Error ? error.message : String(error),
                 errorName: error instanceof Error ? error.name : "UnknownError",
                 errorStack: error instanceof Error ? error.stack : undefined,
@@ -224,9 +225,9 @@ export class NeuroLink {
                     elapsedMs: Date.now() - constructorStartTime,
                     elapsedNs: (process.hrtime.bigint() - constructorHrTimeStart).toString(),
                     memoryInitDurationNs: memoryInitDurationNs.toString(),
-                    memoryInitDurationMs: Number(memoryInitDurationNs) / 1000000,
+                    memoryInitDurationMs: Number(memoryInitDurationNs) / NANOSECOND_TO_MS_DIVISOR,
                     memoryManagerCreateDurationNs: memoryManagerCreateDurationNs.toString(),
-                    memoryManagerCreateDurationMs: Number(memoryManagerCreateDurationNs) / 1000000,
+                    memoryManagerCreateDurationMs: Number(memoryManagerCreateDurationNs) / NANOSECOND_TO_MS_DIVISOR,
                     finalMemoryConfig: {
                         maxSessions: memoryConfig.maxSessions,
                         maxTurnsPerSession: memoryConfig.maxTurnsPerSession,
@@ -245,7 +246,7 @@ export class NeuroLink {
                     elapsedMs: Date.now() - constructorStartTime,
                     elapsedNs: (process.hrtime.bigint() - constructorHrTimeStart).toString(),
                     memoryInitDurationNs: memoryInitDurationNs.toString(),
-                    memoryInitDurationMs: Number(memoryInitDurationNs) / 1000000,
+                    memoryInitDurationMs: Number(memoryInitDurationNs) / NANOSECOND_TO_MS_DIVISOR,
                     error: error instanceof Error ? error.message : String(error),
                     errorName: error instanceof Error ? error.name : "UnknownError",
                     errorStack: error instanceof Error ? error.stack : undefined,
@@ -289,8 +290,8 @@ export class NeuroLink {
             elapsedNs: (process.hrtime.bigint() - constructorHrTimeStart).toString(),
             externalServerInitStartTimeNs: externalServerInitStartTime.toString(),
             serverManagerConfig: {
-                maxServers: 20,
-                defaultTimeout: 15000,
+                maxServers: SERVER_CONFIG.MAX_MCP_SERVERS,
+                defaultTimeout: MCP_TIMEOUTS.EXTERNAL_SERVER_STARTUP_MS,
                 enableAutoRestart: true,
                 enablePerformanceMonitoring: true,
             },
@@ -317,7 +318,7 @@ export class NeuroLink {
                 elapsedMs: Date.now() - constructorStartTime,
                 elapsedNs: (process.hrtime.bigint() - constructorHrTimeStart).toString(),
                 externalServerInitDurationNs: externalServerInitDurationNs.toString(),
-                externalServerInitDurationMs: Number(externalServerInitDurationNs) / 1000000,
+                externalServerInitDurationMs: Number(externalServerInitDurationNs) / NANOSECOND_TO_MS_DIVISOR,
                 hasExternalServerManager: !!this.externalServerManager,
                 message: "External server manager initialized successfully",
             });
@@ -333,7 +334,7 @@ export class NeuroLink {
                 elapsedMs: Date.now() - constructorStartTime,
                 elapsedNs: (process.hrtime.bigint() - constructorHrTimeStart).toString(),
                 externalServerInitDurationNs: externalServerInitDurationNs.toString(),
-                externalServerInitDurationMs: Number(externalServerInitDurationNs) / 1000000,
+                externalServerInitDurationMs: Number(externalServerInitDurationNs) / NANOSECOND_TO_MS_DIVISOR,
                 error: error instanceof Error ? error.message : String(error),
                 errorName: error instanceof Error ? error.name : "UnknownError",
                 errorStack: error instanceof Error ? error.stack : undefined,
@@ -418,7 +419,7 @@ export class NeuroLink {
             elapsedMs: Date.now() - constructorStartTime,
             elapsedNs: (process.hrtime.bigint() - constructorHrTimeStart).toString(),
             eventHandlerSetupDurationNs: eventHandlerSetupDurationNs.toString(),
-            eventHandlerSetupDurationMs: Number(eventHandlerSetupDurationNs) / 1000000,
+            eventHandlerSetupDurationMs: Number(eventHandlerSetupDurationNs) / NANOSECOND_TO_MS_DIVISOR,
             eventHandlersCount: 5,
             eventHandlerTypes: [
                 "connected",
@@ -441,7 +442,7 @@ export class NeuroLink {
             constructorId,
             timestamp: new Date().toISOString(),
             constructorDurationNs: constructorDurationNs.toString(),
-            constructorDurationMs: Number(constructorDurationNs) / 1000000,
+            constructorDurationMs: Number(constructorDurationNs) / NANOSECOND_TO_MS_DIVISOR,
             totalElapsedMs: Date.now() - constructorStartTime,
             finalState: {
                 hasConversationMemory: !!this.conversationMemory,
@@ -542,7 +543,7 @@ export class NeuroLink {
                 elapsedMs: Date.now() - mcpInitStartTime,
                 elapsedNs: (process.hrtime.bigint() - mcpInitHrTimeStart).toString(),
                 performanceImportDurationNs: performanceImportDurationNs.toString(),
-                performanceImportDurationMs: Number(performanceImportDurationNs) / 1000000,
+                performanceImportDurationMs: Number(performanceImportDurationNs) / NANOSECOND_TO_MS_DIVISOR,
                 hasMemoryManager: !!MemoryManager,
                 message: "MemoryManager imported successfully",
             });
@@ -558,7 +559,7 @@ export class NeuroLink {
                 elapsedMs: Date.now() - mcpInitStartTime,
                 elapsedNs: (process.hrtime.bigint() - mcpInitHrTimeStart).toString(),
                 performanceImportDurationNs: performanceImportDurationNs.toString(),
-                performanceImportDurationMs: Number(performanceImportDurationNs) / 1000000,
+                performanceImportDurationMs: Number(performanceImportDurationNs) / NANOSECOND_TO_MS_DIVISOR,
                 error: error instanceof Error ? error.message : String(error),
                 errorName: error instanceof Error ? error.name : "UnknownError",
                 message: "MemoryManager import failed - continuing without performance tracking",
@@ -590,7 +591,7 @@ export class NeuroLink {
      */
     async initializeToolRegistryInternal(mcpInitId, mcpInitStartTime, mcpInitHrTimeStart) {
         const toolRegistryStartTime = process.hrtime.bigint();
-        const initTimeout = 3000;
+        const initTimeout = MCP_TIMEOUTS.INITIALIZATION_MS;
         logger.debug(`[NeuroLink] ⏱️ LOG_POINT_M007_TOOL_REGISTRY_TIMEOUT_SETUP`, {
             logPoint: "M007_TOOL_REGISTRY_TIMEOUT_SETUP",
             mcpInitId,
@@ -616,7 +617,7 @@ export class NeuroLink {
             elapsedMs: Date.now() - mcpInitStartTime,
             elapsedNs: (process.hrtime.bigint() - mcpInitHrTimeStart).toString(),
             toolRegistryDurationNs: toolRegistryDurationNs.toString(),
-            toolRegistryDurationMs: Number(toolRegistryDurationNs) / 1000000,
+            toolRegistryDurationMs: Number(toolRegistryDurationNs) / NANOSECOND_TO_MS_DIVISOR,
             message: "Tool registry initialization completed within timeout",
         });
     }
@@ -644,7 +645,7 @@ export class NeuroLink {
             elapsedMs: Date.now() - mcpInitStartTime,
             elapsedNs: (process.hrtime.bigint() - mcpInitHrTimeStart).toString(),
             providerRegistryDurationNs: providerRegistryDurationNs.toString(),
-            providerRegistryDurationMs: Number(providerRegistryDurationNs) / 1000000,
+            providerRegistryDurationMs: Number(providerRegistryDurationNs) / NANOSECOND_TO_MS_DIVISOR,
             message: "Provider registry registration completed successfully",
         });
     }
@@ -674,7 +675,7 @@ export class NeuroLink {
                 elapsedMs: Date.now() - mcpInitStartTime,
                 elapsedNs: (process.hrtime.bigint() - mcpInitHrTimeStart).toString(),
                 directToolsDurationNs: directToolsDurationNs.toString(),
-                directToolsDurationMs: Number(directToolsDurationNs) / 1000000,
+                directToolsDurationMs: Number(directToolsDurationNs) / NANOSECOND_TO_MS_DIVISOR,
                 serverId: "neurolink-direct",
                 message: "Direct tools server registered successfully",
             });
@@ -692,7 +693,7 @@ export class NeuroLink {
                 elapsedMs: Date.now() - mcpInitStartTime,
                 elapsedNs: (process.hrtime.bigint() - mcpInitHrTimeStart).toString(),
                 directToolsDurationNs: directToolsDurationNs.toString(),
-                directToolsDurationMs: Number(directToolsDurationNs) / 1000000,
+                directToolsDurationMs: Number(directToolsDurationNs) / NANOSECOND_TO_MS_DIVISOR,
                 error: error instanceof Error ? error.message : String(error),
                 errorName: error instanceof Error ? error.name : "UnknownError",
                 errorStack: error instanceof Error ? error.stack : undefined,
@@ -731,7 +732,7 @@ export class NeuroLink {
                 elapsedMs: Date.now() - mcpInitStartTime,
                 elapsedNs: (process.hrtime.bigint() - mcpInitHrTimeStart).toString(),
                 mcpConfigDurationNs: mcpConfigDurationNs.toString(),
-                mcpConfigDurationMs: Number(mcpConfigDurationNs) / 1000000,
+                mcpConfigDurationMs: Number(mcpConfigDurationNs) / NANOSECOND_TO_MS_DIVISOR,
                 serversLoaded: configResult.serversLoaded,
                 errorsCount: configResult.errors.length,
                 configResult: {
@@ -774,7 +775,7 @@ export class NeuroLink {
             initTime: `${initTime}ms`,
             memoryUsed: `${memoryDelta}MB`,
         });
-        if (memoryDelta > 30) {
+        if (memoryDelta > MEMORY_THRESHOLDS.MODERATE_USAGE_MB) {
             mcpLogger.debug("💡 Memory cleanup suggestion: MCP initialization used significant memory. Consider calling MemoryManager.forceGC() after heavy operations.");
         }
     }
@@ -1120,7 +1121,7 @@ export class NeuroLink {
                 elapsedMs: Date.now() - generateInternalStartTime,
                 elapsedNs: (process.hrtime.bigint() - generateInternalHrTimeStart).toString(),
                 conversationMemoryDurationNs: conversationMemoryDurationNs.toString(),
-                conversationMemoryDurationMs: Number(conversationMemoryDurationNs) / 1000000,
+                conversationMemoryDurationMs: Number(conversationMemoryDurationNs) / NANOSECOND_TO_MS_DIVISOR,
                 message: "Conversation memory initialization completed successfully",
             });
         }
@@ -1165,7 +1166,7 @@ export class NeuroLink {
      * Perform MCP generation with retry logic
      */
     async performMCPGenerationRetries(options, generateInternalId, generateInternalStartTime, generateInternalHrTimeStart, functionTag) {
-        const maxMcpRetries = 2;
+        const maxMcpRetries = RETRY_ATTEMPTS.QUICK;
         const mcpRetryLoopStartTime = process.hrtime.bigint();
         logger.debug(`[NeuroLink] 🔄 LOG_POINT_G006_MCP_RETRY_LOOP_START`, {
             logPoint: "G006_MCP_RETRY_LOOP_START",
@@ -1207,7 +1208,7 @@ export class NeuroLink {
                     elapsedMs: Date.now() - generateInternalStartTime,
                     elapsedNs: (process.hrtime.bigint() - generateInternalHrTimeStart).toString(),
                     mcpAttemptDurationNs: mcpAttemptDurationNs.toString(),
-                    mcpAttemptDurationMs: Number(mcpAttemptDurationNs) / 1000000,
+                    mcpAttemptDurationMs: Number(mcpAttemptDurationNs) / NANOSECOND_TO_MS_DIVISOR,
                     currentAttempt: attempt,
                     resultAnalysis: {
                         hasResult: !!mcpResult,
@@ -1326,7 +1327,7 @@ export class NeuroLink {
                 elapsedMs: Date.now() - tryMCPStartTime,
                 elapsedNs: (process.hrtime.bigint() - tryMCPHrTimeStart).toString(),
                 mcpInitCheckDurationNs: mcpInitCheckDurationNs.toString(),
-                mcpInitCheckDurationMs: Number(mcpInitCheckDurationNs) / 1000000,
+                mcpInitCheckDurationMs: Number(mcpInitCheckDurationNs) / NANOSECOND_TO_MS_DIVISOR,
                 mcpInitializedAfter: this.mcpInitialized,
                 initializationSuccessful: this.mcpInitialized,
                 message: "MCP initialization check completed",
@@ -1782,7 +1783,7 @@ export class NeuroLink {
                 elapsedMs: Date.now() - startTime,
                 elapsedNs: (process.hrtime.bigint() - hrTimeStart).toString(),
                 validationDurationNs: validationDurationNs.toString(),
-                validationDurationMs: Number(validationDurationNs) / 1000000,
+                validationDurationMs: Number(validationDurationNs) / NANOSECOND_TO_MS_DIVISOR,
                 validationError: "Stream options must include either input.text or input.audio",
                 message: "EXHAUSTIVE validation failure analysis with character-level debugging",
             });
@@ -1797,7 +1798,7 @@ export class NeuroLink {
             elapsedMs: Date.now() - startTime,
             elapsedNs: (process.hrtime.bigint() - hrTimeStart).toString(),
             validationDurationNs: validationDurationNs.toString(),
-            validationDurationMs: Number(validationDurationNs) / 1000000,
+            validationDurationMs: Number(validationDurationNs) / NANOSECOND_TO_MS_DIVISOR,
             inputTextValid: hasText,
             inputAudioPresent: hasAudio,
             inputTextLength: hasText ? options.input.text.length : 0,
@@ -2315,16 +2316,16 @@ export class NeuroLink {
         this.emitter.emit("tool:start", toolName, params);
         // Set default options
         const finalOptions = {
-            timeout: options?.timeout || 30000, // 30 second default timeout
-            maxRetries: options?.maxRetries || 2, // Default 2 retries for retriable errors
-            retryDelayMs: options?.retryDelayMs || 1000, // 1 second delay between retries
+            timeout: options?.timeout || TOOL_TIMEOUTS.EXECUTION_DEFAULT_MS, // 30 second default timeout
+            maxRetries: options?.maxRetries || RETRY_ATTEMPTS.DEFAULT, // Default 2 retries for retriable errors
+            retryDelayMs: options?.retryDelayMs || RETRY_DELAYS.BASE_MS, // 1 second delay between retries
         };
         // Track memory usage for tool execution
         const { MemoryManager } = await import("./utils/performance.js");
         const startMemory = MemoryManager.getMemoryUsageMB();
         // Get or create circuit breaker for this tool
         if (!this.toolCircuitBreakers.has(toolName)) {
-            this.toolCircuitBreakers.set(toolName, new CircuitBreaker(5, 60000)); // 5 failures, 1 minute timeout
+            this.toolCircuitBreakers.set(toolName, new CircuitBreaker(CIRCUIT_BREAKER.FAILURE_THRESHOLD, CIRCUIT_BREAKER_RESET_MS));
         }
         const circuitBreaker = this.toolCircuitBreakers.get(toolName);
         // Initialize metrics for this tool if not exists
@@ -2647,10 +2648,10 @@ export class NeuroLink {
             // Check memory usage after tool enumeration
             const endMemory = MemoryManager.getMemoryUsageMB();
             const memoryDelta = endMemory.heapUsed - startMemory.heapUsed;
-            if (memoryDelta > 10) {
+            if (memoryDelta > MEMORY_THRESHOLDS.LOW_USAGE_MB) {
                 mcpLogger.debug(`🔍 Tool listing used ${memoryDelta}MB memory (large tool registry detected)`);
                 // Optimized collection patterns should reduce memory usage significantly
-                if (uniqueTools.length > 100) {
+                if (uniqueTools.length > PERFORMANCE_THRESHOLDS.LARGE_TOOL_COLLECTION) {
                     mcpLogger.debug("💡 Tool collection optimized for large sets. Memory usage reduced through efficient object reuse.");
                 }
             }
@@ -2722,7 +2723,7 @@ export class NeuroLink {
                     try {
                         const response = await fetch("http://localhost:11434/api/tags", {
                             method: "GET",
-                            signal: AbortSignal.timeout(2000),
+                            signal: AbortSignal.timeout(PROVIDER_TIMEOUTS.AUTH_MS),
                         });
                         if (!response.ok) {
                             throw new Error("Ollama service not responding");
@@ -3099,7 +3100,7 @@ export class NeuroLink {
     resetToolCircuitBreaker(toolName) {
         if (this.toolCircuitBreakers.has(toolName)) {
             // Create a new circuit breaker (effectively resets it)
-            this.toolCircuitBreakers.set(toolName, new CircuitBreaker(5, 60000));
+            this.toolCircuitBreakers.set(toolName, new CircuitBreaker(CIRCUIT_BREAKER.FAILURE_THRESHOLD, CIRCUIT_BREAKER_RESET_MS));
             mcpLogger.info(`Circuit breaker reset for tool: ${toolName}`);
         }
     }

package/dist/lib/providers/amazonBedrock.js

@@ -157,7 +157,7 @@ export class AmazonBedrockProvider extends BaseProvider {
                     },
                 ],
                 inferenceConfig: {
-                    maxTokens: options.maxTokens || 4096,
+                    maxTokens: options.maxTokens, // No default limit - unlimited unless specified
                     temperature: options.temperature || 0.7,
                 },
             };
@@ -718,7 +718,7 @@ export class AmazonBedrockProvider extends BaseProvider {
                 },
             ],
             inferenceConfig: {
-                maxTokens: options.maxTokens || 4096,
+                maxTokens: options.maxTokens, // No default limit - unlimited unless specified
                 temperature: options.temperature || 0.7,
             },
         };

package/dist/lib/providers/anthropic.js

@@ -5,7 +5,7 @@ import { BaseProvider } from "../core/baseProvider.js";
 import { logger } from "../utils/logger.js";
 import { createTimeoutController, TimeoutError } from "../utils/timeout.js";
 import { AuthenticationError, NetworkError, ProviderError, RateLimitError, } from "../types/errors.js";
-import { DEFAULT_MAX_TOKENS, DEFAULT_MAX_STEPS } from "../core/constants.js";
+import { DEFAULT_MAX_STEPS } from "../core/constants.js";
 import { validateApiKey, createAnthropicConfig, getProviderModel, } from "../utils/providerConfig.js";
 import { buildMessagesArray } from "../utils/messageBuilder.js";
 import { createProxyFetch } from "../proxy/proxyFetch.js";
@@ -98,7 +98,7 @@ export class AnthropicProvider extends BaseProvider {
                 model: this.model,
                 messages: messages,
                 temperature: options.temperature,
-                maxTokens: options.maxTokens || DEFAULT_MAX_TOKENS,
+                maxTokens: options.maxTokens, // No default limit - unlimited unless specified
                 tools,
                 maxSteps: options.maxSteps || DEFAULT_MAX_STEPS,
                 toolChoice: shouldUseTools ? "auto" : "none",
@@ -111,22 +111,13 @@ export class AnthropicProvider extends BaseProvider {
             // Full tool support is now available with real streaming
             const toolCalls = [];
             const toolResults = [];
-            const usage = await result.usage;
-            const finishReason = await result.finishReason;
             return {
                 stream: transformedStream,
                 provider: this.providerName,
                 model: this.modelName,
                 toolCalls, // ✅ Include tool calls in stream result
                 toolResults, // ✅ Include tool results in stream result
-                usage: usage
-                    ? {
-                        input: usage.promptTokens || 0,
-                        output: usage.completionTokens || 0,
-                        total: usage.totalTokens || 0,
-                    }
-                    : undefined,
-                finishReason: finishReason || undefined,
+                // Note: omit usage/finishReason to avoid blocking streaming; compute asynchronously if needed.
             };
         }
         catch (error) {

package/dist/lib/providers/anthropicBaseProvider.js

@@ -4,7 +4,6 @@ import { AnthropicModels } from "../types/index.js";
 import { BaseProvider } from "../core/baseProvider.js";
 import { logger } from "../utils/logger.js";
 import { createTimeoutController, TimeoutError } from "../utils/timeout.js";
-import { DEFAULT_MAX_TOKENS } from "../core/constants.js";
 import { validateApiKey, createAnthropicBaseConfig, } from "../utils/providerConfig.js";
 /**
  * Anthropic provider implementation using BaseProvider pattern
@@ -70,7 +69,7 @@ export class AnthropicProviderV2 extends BaseProvider {
                 prompt: options.input.text,
                 system: options.systemPrompt,
                 temperature: options.temperature,
-                maxTokens: options.maxTokens || DEFAULT_MAX_TOKENS,
+                maxTokens: options.maxTokens, // No default limit - unlimited unless specified
                 tools: options.tools,
                 toolChoice: "auto",
                 abortSignal: timeoutController?.controller.signal,

package/dist/lib/providers/azureOpenai.d.ts

@@ -1,6 +1,6 @@
 import { type LanguageModelV1 } from "ai";
 import { BaseProvider } from "../core/baseProvider.js";
-import type { AIProviderName } from "../types/index.js";
+import type { AIProviderName } from "../core/types.js";
 import type { StreamOptions, StreamResult } from "../types/streamTypes.js";
 export declare class AzureOpenAIProvider extends BaseProvider {
     private apiKey;

package/dist/lib/providers/azureOpenai.js

@@ -1,10 +1,12 @@
 import { createAzure } from "@ai-sdk/azure";
 import { streamText } from "ai";
 import { BaseProvider } from "../core/baseProvider.js";
+import { APIVersions } from "../core/types.js";
 import { validateApiKey, createAzureAPIKeyConfig, createAzureEndpointConfig, } from "../utils/providerConfig.js";
 import { logger } from "../utils/logger.js";
 import { buildMessagesArray } from "../utils/messageBuilder.js";
 import { createProxyFetch } from "../proxy/proxyFetch.js";
+import { DEFAULT_MAX_STEPS } from "../core/constants.js";
 export class AzureOpenAIProvider extends BaseProvider {
     apiKey;
     resourceName;
@@ -18,13 +20,15 @@ export class AzureOpenAIProvider extends BaseProvider {
         this.resourceName = endpoint
             .replace("https://", "")
             .replace(/\/+$/, "") // Remove trailing slashes
-            .replace(".openai.azure.com", "");
+            .replace(".openai.azure.com", "")
+            .replace(".cognitiveservices.azure.com", "");
         this.deployment =
             modelName ||
+                process.env.AZURE_OPENAI_MODEL ||
                 process.env.AZURE_OPENAI_DEPLOYMENT ||
                 process.env.AZURE_OPENAI_DEPLOYMENT_ID ||
                 "gpt-4o";
-        this.apiVersion = process.env.AZURE_API_VERSION || "2024-10-01-preview";
+        this.apiVersion = process.env.AZURE_API_VERSION || APIVersions.AZURE_LATEST;
         // Configuration validation - now using consolidated utility
         if (!this.apiKey) {
             validateApiKey(createAzureAPIKeyConfig());
@@ -33,6 +37,7 @@ export class AzureOpenAIProvider extends BaseProvider {
             validateApiKey(createAzureEndpointConfig());
         }
         // Create the Azure provider instance with proxy support
+        // Let the Azure SDK handle all URL construction automatically
         this.azureProvider = createAzure({
             resourceName: this.resourceName,
             apiKey: this.apiKey,
@@ -72,20 +77,57 @@ export class AzureOpenAIProvider extends BaseProvider {
     // executeGenerate removed - BaseProvider handles all generation with tools
     async executeStream(options, _analysisSchema) {
         try {
+            // Get ALL available tools (direct + MCP + external from options) - EXACTLY like BaseProvider
+            const shouldUseTools = !options.disableTools && this.supportsTools();
+            const baseTools = shouldUseTools ? await this.getAllTools() : {};
+            const tools = shouldUseTools
+                ? {
+                    ...baseTools,
+                    ...(options.tools || {}), // Include external tools passed from NeuroLink
+                }
+                : undefined;
+            // DEBUG: Log detailed tool information
+            logger.debug("Azure Stream - Tool Loading Debug", {
+                shouldUseTools,
+                baseToolsProvided: !!baseTools,
+                baseToolCount: baseTools ? Object.keys(baseTools).length : 0,
+                finalToolCount: tools ? Object.keys(tools).length : 0,
+                toolNames: tools ? Object.keys(tools).slice(0, 10) : [],
+                disableTools: options.disableTools,
+                supportsTools: this.supportsTools(),
+                externalToolsCount: options.tools
+                    ? Object.keys(options.tools).length
+                    : 0,
+            });
+            if (tools && Object.keys(tools).length > 0) {
+                logger.debug("Azure Stream - First 5 Tools Detail", {
+                    tools: Object.keys(tools)
+                        .slice(0, 5)
+                        .map((name) => ({
+                        name,
+                        description: tools[name]?.description?.substring(0, 100),
+                    })),
+                });
+            }
             // Build message array from options
             const messages = buildMessagesArray(options);
             const stream = await streamText({
                 model: this.azureProvider(this.deployment),
                 messages: messages,
-                maxTokens: options.maxTokens || 1000,
-                temperature: options.temperature || 0.7,
+                ...(options.maxTokens !== null && options.maxTokens !== undefined
+                    ? { maxTokens: options.maxTokens }
+                    : {}),
+                ...(options.temperature !== null && options.temperature !== undefined
+                    ? { temperature: options.temperature }
+                    : {}),
+                tools,
+                toolChoice: shouldUseTools ? "auto" : "none",
+                maxSteps: options.maxSteps || DEFAULT_MAX_STEPS,
             });
+            // Transform string stream to content object stream using BaseProvider method
+            const transformedStream = this.createTextStream(stream);
             return {
-                stream: (async function* () {
-                    for await (const chunk of stream.textStream) {
-                        yield { content: chunk };
-                    }
-                })(),
+                stream: transformedStream,
                 provider: "azure",
                 model: this.deployment,
                 metadata: {

package/dist/lib/providers/googleAiStudio.js

@@ -5,12 +5,12 @@ import { BaseProvider } from "../core/baseProvider.js";
 import { logger } from "../utils/logger.js";
 import { createTimeoutController, TimeoutError } from "../utils/timeout.js";
 import { AuthenticationError, NetworkError, ProviderError, RateLimitError, } from "../types/errors.js";
-import { DEFAULT_MAX_TOKENS, DEFAULT_MAX_STEPS } from "../core/constants.js";
+import { DEFAULT_MAX_STEPS } from "../core/constants.js";
 import { streamAnalyticsCollector } from "../core/streamAnalytics.js";
 import { buildMessagesArray } from "../utils/messageBuilder.js";
 // Create Google GenAI client
 async function createGoogleGenAIClient(apiKey) {
-    const mod = await import("@google/genai");
+    const mod = await import("@google/generative-ai");
     const ctor = mod.GoogleGenAI;
     if (!ctor) {
         throw new Error("@google/genai does not export GoogleGenAI");
@@ -96,7 +96,7 @@ export class GoogleAIStudioProvider extends BaseProvider {
                 model,
                 messages: messages,
                 temperature: options.temperature,
-                maxTokens: options.maxTokens || DEFAULT_MAX_TOKENS,
+                maxTokens: options.maxTokens, // No default limit - unlimited unless specified
                 tools,
                 maxSteps: options.maxSteps || DEFAULT_MAX_STEPS,
                 toolChoice: shouldUseTools ? "auto" : "none",

package/dist/lib/providers/googleVertex.js

@@ -4,7 +4,7 @@ import { streamText, Output, } from "ai";
 import { BaseProvider } from "../core/baseProvider.js";
 import { logger } from "../utils/logger.js";
 import { createTimeoutController, TimeoutError } from "../utils/timeout.js";
-import { DEFAULT_MAX_TOKENS, DEFAULT_MAX_STEPS } from "../core/constants.js";
+import { DEFAULT_MAX_STEPS } from "../core/constants.js";
 import { ModelConfigurationManager } from "../core/modelConfiguration.js";
 import { validateApiKey, createVertexProjectConfig, createGoogleAuthConfig, } from "../utils/providerConfig.js";
 import fs from "fs";
@@ -945,7 +945,7 @@ export class GoogleVertexProvider extends BaseProvider {
             // This avoids hardcoded model-specific logic and repeated config lookups
             const shouldSetMaxTokens = this.shouldSetMaxTokensCached(modelName);
             const maxTokens = shouldSetMaxTokens
-                ? options.maxTokens || DEFAULT_MAX_TOKENS
+                ? options.maxTokens // No default limit
                 : undefined;
             // Build complete stream options with proper typing
             let streamOptions = {

package/dist/lib/providers/huggingFace.js

@@ -3,7 +3,6 @@ import { streamText, } from "ai";
 import { BaseProvider } from "../core/baseProvider.js";
 import { logger } from "../utils/logger.js";
 import { createTimeoutController, TimeoutError } from "../utils/timeout.js";
-import { DEFAULT_MAX_TOKENS } from "../core/constants.js";
 import { validateApiKey, createHuggingFaceConfig, getProviderModel, } from "../utils/providerConfig.js";
 import { buildMessagesArray } from "../utils/messageBuilder.js";
 import { createProxyFetch } from "../proxy/proxyFetch.js";
@@ -120,7 +119,7 @@ export class HuggingFaceProvider extends BaseProvider {
                 model: this.model,
                 messages: messages,
                 temperature: options.temperature,
-                maxTokens: options.maxTokens || DEFAULT_MAX_TOKENS,
+                maxTokens: options.maxTokens, // No default limit - unlimited unless specified
                 tools: streamOptions.tools, // Tools format conversion handled by prepareStreamOptions
                 toolChoice: streamOptions.toolChoice, // Tool choice handled by prepareStreamOptions
                 abortSignal: timeoutController?.controller.signal,

package/dist/lib/providers/litellm.js

@@ -3,7 +3,6 @@ import { streamText } from "ai";
 import { BaseProvider } from "../core/baseProvider.js";
 import { logger } from "../utils/logger.js";
 import { createTimeoutController, TimeoutError } from "../utils/timeout.js";
-import { DEFAULT_MAX_TOKENS } from "../core/constants.js";
 import { getProviderModel } from "../utils/providerConfig.js";
 import { streamAnalyticsCollector } from "../core/streamAnalytics.js";
 import { buildMessagesArray } from "../utils/messageBuilder.js";
@@ -127,7 +126,7 @@ export class LiteLLMProvider extends BaseProvider {
                 model: this.model,
                 messages: messages,
                 temperature: options.temperature,
-                maxTokens: options.maxTokens || DEFAULT_MAX_TOKENS,
+                maxTokens: options.maxTokens, // No default limit - unlimited unless specified
                 tools: options.tools,
                 toolChoice: "auto",
                 abortSignal: timeoutController?.controller.signal,

package/dist/lib/providers/mistral.js

@@ -3,7 +3,7 @@ import { streamText } from "ai";
 import { BaseProvider } from "../core/baseProvider.js";
 import { logger } from "../utils/logger.js";
 import { createTimeoutController, TimeoutError } from "../utils/timeout.js";
-import { DEFAULT_MAX_TOKENS, DEFAULT_MAX_STEPS } from "../core/constants.js";
+import { DEFAULT_MAX_STEPS } from "../core/constants.js";
 import { validateApiKey, createMistralConfig, getProviderModel, } from "../utils/providerConfig.js";
 import { streamAnalyticsCollector } from "../core/streamAnalytics.js";
 import { buildMessagesArray } from "../utils/messageBuilder.js";
@@ -54,7 +54,7 @@ export class MistralProvider extends BaseProvider {
                 model: this.model,
                 messages: messages,
                 temperature: options.temperature,
-                maxTokens: options.maxTokens || DEFAULT_MAX_TOKENS,
+                maxTokens: options.maxTokens, // No default limit - unlimited unless specified
                 tools,
                 maxSteps: options.maxSteps || DEFAULT_MAX_STEPS,
                 toolChoice: shouldUseTools ? "auto" : "none",

package/dist/lib/providers/ollama.js

@@ -1,6 +1,5 @@
 import { BaseProvider } from "../core/baseProvider.js";
 import { logger } from "../utils/logger.js";
-import { DEFAULT_MAX_TOKENS } from "../core/constants.js";
 import { modelConfig } from "../core/modelConfiguration.js";
 import { createProxyFetch } from "../proxy/proxyFetch.js";
 import { TimeoutError } from "../utils/timeout.js";
@@ -88,10 +87,10 @@ class OllamaLanguageModel {
         return {
             text: data.response,
             usage: {
-                promptTokens: data.prompt_eval_count || this.estimateTokens(prompt),
-                completionTokens: data.eval_count || this.estimateTokens(data.response),
-                totalTokens: (data.prompt_eval_count || this.estimateTokens(prompt)) +
-                    (data.eval_count || this.estimateTokens(data.response)),
+                promptTokens: data.prompt_eval_count ?? this.estimateTokens(prompt),
+                completionTokens: data.eval_count ?? this.estimateTokens(String(data.response ?? "")),
+                totalTokens: (data.prompt_eval_count ?? this.estimateTokens(prompt)) +
+                    (data.eval_count ?? this.estimateTokens(String(data.response ?? ""))),
             },
             finishReason: "stop",
             rawCall: {
@@ -271,7 +270,7 @@ export class OllamaProvider extends BaseProvider {
      * @returns true for supported models, false for unsupported models
      */
     supportsTools() {
-        const modelName = this.modelName.toLowerCase();
+        const modelName = (this.modelName ?? getDefaultOllamaModel()).toLowerCase();
         // Get tool-capable models from configuration
         const ollamaConfig = modelConfig.getProviderConfiguration("ollama");
         const toolCapableModels = ollamaConfig?.modelBehavior?.toolCapableModels || [];
@@ -340,7 +339,7 @@ export class OllamaProvider extends BaseProvider {
                 tool_choice: "auto",
                 stream: true,
                 temperature: options.temperature,
-                max_tokens: options.maxTokens || DEFAULT_MAX_TOKENS,
+                max_tokens: options.maxTokens,
             }),
             signal: createAbortSignalWithTimeout(this.timeout),
         });
@@ -381,7 +380,7 @@ export class OllamaProvider extends BaseProvider {
                 stream: true,
                 options: {
                     temperature: options.temperature,
-                    num_predict: options.maxTokens || DEFAULT_MAX_TOKENS,
+                    num_predict: options.maxTokens,
                 },
             }),
             signal: createAbortSignalWithTimeout(this.timeout),

package/dist/lib/providers/openAI.js

@@ -5,7 +5,7 @@ import { BaseProvider } from "../core/baseProvider.js";
 import { logger } from "../utils/logger.js";
 import { createTimeoutController, TimeoutError } from "../utils/timeout.js";
 import { AuthenticationError, InvalidModelError, NetworkError, ProviderError, RateLimitError, } from "../types/errors.js";
-import { DEFAULT_MAX_TOKENS, DEFAULT_MAX_STEPS } from "../core/constants.js";
+import { DEFAULT_MAX_STEPS } from "../core/constants.js";
 import { validateApiKey, createOpenAIConfig, getProviderModel, } from "../utils/providerConfig.js";
 import { streamAnalyticsCollector } from "../core/streamAnalytics.js";
 import { buildMessagesArray } from "../utils/messageBuilder.js";
@@ -97,7 +97,7 @@ export class OpenAIProvider extends BaseProvider {
                 model: this.model,
                 messages: messages,
                 temperature: options.temperature,
-                maxTokens: options.maxTokens || DEFAULT_MAX_TOKENS,
+                maxTokens: options.maxTokens, // No default limit - unlimited unless specified
                 tools,
                 maxSteps: options.maxSteps || DEFAULT_MAX_STEPS,
                 toolChoice: shouldUseTools ? "auto" : "none",