snow-flow 2.9.8 → 2.9.9

package/dist/mcp/servicenow-machine-learning-mcp.js

@@ -38,6 +38,12 @@ var __importStar = (this && this.__importStar) || (function () {
 })();
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.ServiceNowMachineLearningMCP = void 0;
+// CRITICAL FIX: Add performance polyfill for TensorFlow.js in Node.js environment
+// This fixes the "Cannot read properties of undefined (reading 'tick')" error
+if (typeof global !== 'undefined' && !global.performance) {
+    const { performance } = require('perf_hooks');
+    global.performance = performance;
+}
 const index_js_1 = require("@modelcontextprotocol/sdk/server/index.js");
 const stdio_js_1 = require("@modelcontextprotocol/sdk/server/stdio.js");
 const types_js_1 = require("@modelcontextprotocol/sdk/types.js");
@@ -766,14 +772,22 @@ class ServiceNowMachineLearningMCP {
                 metrics: ['accuracy']
             });
             this.logger.info('Training incident classifier...');
-            // Train model
+            // Train model with improved error handling
             const history = await model.fit(features, labels, {
                 epochs,
                 validationSplit: validation_split,
                 batchSize: 32,
                 callbacks: {
                     onEpochEnd: (epoch, logs) => {
-                        this.logger.info(`Epoch ${epoch + 1}: loss = ${logs?.loss?.toFixed(4)}, accuracy = ${logs?.acc?.toFixed(4)}`);
+                        try {
+                            const loss = logs?.loss ? logs.loss.toFixed(4) : 'N/A';
+                            const accuracy = logs?.acc ? logs.acc.toFixed(4) : 'N/A';
+                            this.logger.info(`Epoch ${epoch + 1}: loss = ${loss}, accuracy = ${accuracy}`);
+                        }
+                        catch (e) {
+                            // Ignore callback errors to prevent training interruption
+                            this.logger.warn(`Callback error in epoch ${epoch + 1}:`, e);
+                        }
                     }
                 }
             });
@@ -849,15 +863,21 @@ class ServiceNowMachineLearningMCP {
                 loss: 'categoricalCrossentropy',
                 metrics: ['accuracy']
             });
-            // Train
+            // Train with improved error handling
             const history = await model.fit(features, labels, {
                 epochs: 100,
                 validationSplit: 0.2,
                 batchSize: 16,
                 callbacks: {
                     onEpochEnd: (epoch, logs) => {
-                        if (epoch % 10 === 0) {
-                            this.logger.info(`Epoch ${epoch}: accuracy = ${logs?.acc?.toFixed(4)}`);
+                        try {
+                            if (epoch % 10 === 0) {
+                                const accuracy = logs?.acc ? logs.acc.toFixed(4) : 'N/A';
+                                this.logger.info(`Epoch ${epoch}: accuracy = ${accuracy}`);
+                            }
+                        }
+                        catch (e) {
+                            this.logger.warn(`Callback error in epoch ${epoch}:`, e);
                         }
                     }
                 }
@@ -948,15 +968,21 @@ class ServiceNowMachineLearningMCP {
                 optimizer: tf.train.adam(0.001),
                 loss: 'meanSquaredError'
             });
-            // Train
+            // Train with improved error handling
             await autoencoder.fit(normalized, normalized, {
                 epochs: 100,
                 batchSize: 32,
                 validationSplit: 0.1,
                 callbacks: {
                     onEpochEnd: (epoch, logs) => {
-                        if (epoch % 20 === 0) {
-                            this.logger.info(`Anomaly detector epoch ${epoch}: loss = ${logs?.loss?.toFixed(6)}`);
+                        try {
+                            if (epoch % 20 === 0) {
+                                const loss = logs?.loss ? logs.loss.toFixed(6) : 'N/A';
+                                this.logger.info(`Anomaly detector epoch ${epoch}: loss = ${loss}`);
+                            }
+                        }
+                        catch (e) {
+                            this.logger.warn(`Callback error in epoch ${epoch}:`, e);
                         }
                     }
                 }
@@ -1769,15 +1795,21 @@ class ServiceNowMachineLearningMCP {
             batchIncidents.forEach(inc => allCategories.add(inc.category));
             // Process batch with feature hashing
             const { features, labels } = this.processBatchWithHashing(batchIncidents, Array.from(allCategories), featureHasher);
-            // Train on batch
+            // Train on batch with improved error handling
             await model.fit(features, labels, {
                 epochs: Math.ceil(epochs / totalBatches), // Distribute epochs across batches
                 batchSize: 32,
                 verbose: 0,
                 callbacks: {
                     onBatchEnd: async (batch, logs) => {
-                        if (batch % 10 === 0) {
-                            this.logger.info(`Batch ${batch}: loss=${logs?.loss?.toFixed(4)}`);
+                        try {
+                            if (batch % 10 === 0 && logs?.loss) {
+                                this.logger.info(`Batch ${batch}: loss=${logs.loss.toFixed(4)}`);
+                            }
+                        }
+                        catch (e) {
+                            // Ignore callback errors to prevent training interruption
+                            this.logger.warn(`Callback error in batch ${batch}:`, e);
                         }
                     }
                 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "snow-flow",
-  "version": "2.9.8",
+  "version": "2.9.9",
   "description": "Snow-Flow: ServiceNow Advanced Intelligence Platform - 100+ real MCP tools with AI-powered swarm orchestration and neural networks. Dynamic task categorization using AI. Machine learning for incident classification, change risk prediction, and anomaly detection. Zero Mock Data, 100% Real API Integration.",
   "main": "dist/index.js",
   "type": "commonjs",