@juspay/neurolink 7.33.3 → 7.33.4

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",

package/dist/lib/providers/openaiCompatible.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 { streamAnalyticsCollector } from "../core/streamAnalytics.js";
 import { createProxyFetch } from "../proxy/proxyFetch.js";
 // Constants
@@ -163,7 +162,7 @@ export class OpenAICompatibleProvider 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,
@@ -207,12 +206,16 @@ export class OpenAICompatibleProvider extends BaseProvider {
             const modelsUrl = new URL("/v1/models", this.config.baseURL).toString();
             logger.debug(`Fetching available models from: ${modelsUrl}`);
             const proxyFetch = createProxyFetch();
+            const controller = new AbortController();
+            const t = setTimeout(() => controller.abort(), 5000);
             const response = await proxyFetch(modelsUrl, {
                 headers: {
                     Authorization: `Bearer ${this.config.apiKey}`,
                     "Content-Type": "application/json",
                 },
+                signal: controller.signal,
             });
+            clearTimeout(t);
             if (!response.ok) {
                 logger.warn(`Models endpoint returned ${response.status}: ${response.statusText}`);
                 return this.getFallbackModels();

package/dist/lib/providers/sagemaker/language-model.d.ts

@@ -9,6 +9,11 @@ import type { SageMakerConfig, SageMakerModelConfig } from "./types.js";
 import type { ConnectivityResult } from "../../types/typeAliases.js";
 /**
  * SageMaker Language Model implementing LanguageModelV1 interface
+ *
+ * Token Limit Behavior:
+ * - When maxTokens is undefined, SageMaker uses the model's default token limits
+ * - When maxTokens is specified, it sets max_new_tokens parameter explicitly
+ * - This aligns with the unlimited-by-default token policy across all providers
  */
 export declare class SageMakerLanguageModel implements LanguageModelV1 {
     readonly specificationVersion = "v1";

package/dist/lib/providers/sagemaker/language-model.js

@@ -75,6 +75,11 @@ const DEFAULT_MAX_CONCURRENCY = 10;
 const DEFAULT_MIN_CONCURRENCY = 1;
 /**
  * SageMaker Language Model implementing LanguageModelV1 interface
+ *
+ * Token Limit Behavior:
+ * - When maxTokens is undefined, SageMaker uses the model's default token limits
+ * - When maxTokens is specified, it sets max_new_tokens parameter explicitly
+ * - This aligns with the unlimited-by-default token policy across all providers
  */
 export class SageMakerLanguageModel {
     specificationVersion = "v1";
@@ -345,7 +350,10 @@ export class SageMakerLanguageModel {
         const request = {
             inputs: promptText,
             parameters: {
-                max_new_tokens: options.maxTokens || 512,
+                // Only include max_new_tokens if explicitly specified; let SageMaker use model defaults otherwise
+                ...(options.maxTokens !== undefined
+                    ? { max_new_tokens: options.maxTokens }
+                    : {}),
                 temperature: options.temperature || 0.7,
                 top_p: options.topP || 0.9,
                 stop: options.stopSequences || [],

package/dist/lib/utils/providerHealth.js

@@ -639,9 +639,13 @@ export class ProviderHealthChecker {
             healthStatus.configurationIssues.push("Invalid AZURE_OPENAI_ENDPOINT format");
             healthStatus.recommendations.push("Set AZURE_OPENAI_ENDPOINT to a valid URL (e.g., https://your-resource.openai.azure.com/)");
         }
-        if (!process.env.AZURE_OPENAI_DEPLOYMENT_NAME) {
-            healthStatus.configurationIssues.push("AZURE_OPENAI_DEPLOYMENT_NAME not set");
-            healthStatus.recommendations.push("Set AZURE_OPENAI_DEPLOYMENT_NAME to your deployment name");
+        // Check for deployment name using the SAME logic as the Azure provider
+        const deploymentName = process.env.AZURE_OPENAI_MODEL ||
+            process.env.AZURE_OPENAI_DEPLOYMENT ||
+            process.env.AZURE_OPENAI_DEPLOYMENT_ID;
+        if (!deploymentName) {
+            healthStatus.configurationIssues.push("No Azure deployment specified");
+            healthStatus.recommendations.push("Set one of: AZURE_OPENAI_MODEL, AZURE_OPENAI_DEPLOYMENT, or AZURE_OPENAI_DEPLOYMENT_ID");
         }
     }
     /**

package/dist/lib/utils/tokenLimits.d.ts

@@ -6,13 +6,13 @@ import { PROVIDER_MAX_TOKENS } from "../core/constants.js";
 /**
  * Get the safe maximum tokens for a provider and model
  */
-export declare function getSafeMaxTokens(provider: keyof typeof PROVIDER_MAX_TOKENS | string, model?: string, requestedMaxTokens?: number): number;
+export declare function getSafeMaxTokens(provider: keyof typeof PROVIDER_MAX_TOKENS | string, model?: string, requestedMaxTokens?: number): number | undefined;
 /**
  * Validate if maxTokens is safe for a provider/model combination
  */
 export declare function validateMaxTokens(provider: keyof typeof PROVIDER_MAX_TOKENS | string, model?: string, maxTokens?: number): {
     isValid: boolean;
-    recommendedMaxTokens: number;
+    recommendedMaxTokens?: number;
     warning?: string;
 };
 /**

package/dist/lib/utils/tokenLimits.js

@@ -2,7 +2,7 @@
  * Provider-specific token limit utilities
  * Provides safe maxTokens values based on provider and model capabilities
  */
-import { PROVIDER_MAX_TOKENS, DEFAULT_MAX_TOKENS } from "../core/constants.js";
+import { PROVIDER_MAX_TOKENS } from "../core/constants.js";
 import { logger } from "./logger.js";
 /**
  * Get the safe maximum tokens for a provider and model
@@ -11,8 +11,8 @@ export function getSafeMaxTokens(provider, model, requestedMaxTokens) {
     // Get provider-specific limits
     const providerLimits = PROVIDER_MAX_TOKENS[provider];
     if (!providerLimits) {
-        logger.warn(`Unknown provider ${provider}, using default maxTokens limit`);
-        return Math.min(requestedMaxTokens || DEFAULT_MAX_TOKENS, PROVIDER_MAX_TOKENS.default);
+        logger.warn(`Unknown provider ${provider}, no token limits enforced`);
+        return requestedMaxTokens || undefined; // No default limit for unknown providers
     }
     // Get model-specific limit or provider default
     let maxLimit;
@@ -54,6 +54,13 @@ export function validateMaxTokens(provider, model, maxTokens) {
             recommendedMaxTokens: safeMaxTokens,
         };
     }
+    // If no limits are defined, validation always passes
+    if (safeMaxTokens === undefined) {
+        return {
+            isValid: true,
+            recommendedMaxTokens: maxTokens,
+        };
+    }
     const isValid = maxTokens <= safeMaxTokens;
     return {
         isValid,

package/dist/mcp/servers/aiProviders/aiWorkflowTools.js

@@ -146,7 +146,7 @@ Return ONLY a valid JSON object with this exact structure:
 Generate 3-5 comprehensive test cases covering the requested types.`;
             const result = await provider.generate({
                 prompt: prompt,
-                maxTokens: Math.floor(DEFAULT_MAX_TOKENS * 1.2),
+                maxTokens: 10000, // High limit for complex analysis
                 temperature: 0.3, // Lower temperature for more consistent structured output
             });
             if (!result || !result.content) {
@@ -360,7 +360,7 @@ Return ONLY a valid JSON object with this exact structure:
 Focus on creating accurate, useful documentation that explains the code's purpose, parameters, return values, and usage patterns.`;
             const result = await provider.generate({
                 prompt: prompt,
-                maxTokens: Math.floor(DEFAULT_MAX_TOKENS * 1.2),
+                maxTokens: 10000, // High limit for complex analysis
                 temperature: 0.3, // Moderate temperature for creative but structured documentation
             });
             if (!result || !result.content) {

package/dist/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/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/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/providers/azureOpenai.js

@@ -6,6 +6,7 @@ import { validateApiKey, createAzureAPIKeyConfig, createAzureEndpointConfig, } f
 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;
@@ -19,9 +20,11 @@ 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";
@@ -34,6 +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,
@@ -73,20 +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/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/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/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/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/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/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),