agentic-qe 2.5.6 → 2.5.7

package/dist/agents/n8n/N8nFailureModeTesterAgent.js

@@ -0,0 +1,956 @@
+"use strict";
+/**
+ * N8nFailureModeTesterAgent
+ *
+ * Error handling and retry semantics testing for n8n workflows:
+ * - Retry/backoff behavior validation
+ * - Partial failure handling in loops
+ * - Error branch testing
+ * - "Continue on fail" behavior
+ * - Dead-letter patterns
+ * - Timeout handling
+ * - Error propagation analysis
+ */
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.N8nFailureModeTesterAgent = void 0;
+const N8nBaseAgent_1 = require("./N8nBaseAgent");
+const N8nTestHarness_1 = require("./N8nTestHarness");
+// ============================================================================
+// Error Handling Patterns
+// ============================================================================
+const ERROR_PRONE_NODE_TYPES = [
+    'n8n-nodes-base.httpRequest',
+    'n8n-nodes-base.webhook',
+    'n8n-nodes-base.executeCommand',
+    'n8n-nodes-base.function',
+    'n8n-nodes-base.code',
+    'n8n-nodes-base.postgres',
+    'n8n-nodes-base.mysql',
+    'n8n-nodes-base.mongodb',
+    'n8n-nodes-base.redis',
+    'n8n-nodes-base.ftp',
+    'n8n-nodes-base.ssh',
+    'n8n-nodes-base.awsLambda',
+];
+const LOOP_NODE_TYPES = [
+    'n8n-nodes-base.splitInBatches',
+    'n8n-nodes-base.loop',
+];
+// ============================================================================
+// Agent Implementation
+// ============================================================================
+class N8nFailureModeTesterAgent extends N8nBaseAgent_1.N8nBaseAgent {
+    constructor(config) {
+        const capabilities = [
+            {
+                name: 'retry-analysis',
+                version: '1.0.0',
+                description: 'Analyze retry configurations and backoff behavior',
+                parameters: {},
+            },
+            {
+                name: 'error-branch-testing',
+                version: '1.0.0',
+                description: 'Test error branch handling',
+                parameters: {},
+            },
+            {
+                name: 'continue-on-fail-analysis',
+                version: '1.0.0',
+                description: 'Analyze continue-on-fail behavior and risks',
+                parameters: {},
+            },
+            {
+                name: 'fault-injection',
+                version: '1.0.0',
+                description: 'Inject faults to test error handling',
+                parameters: {},
+            },
+        ];
+        super({
+            ...config,
+            type: 'n8n-failure-mode-tester',
+            capabilities: [...capabilities, ...(config.capabilities || [])],
+        });
+    }
+    async performTask(task) {
+        const failureTask = task;
+        if (failureTask.type !== 'failure-mode-test') {
+            throw new Error(`Unsupported task type: ${failureTask.type}`);
+        }
+        return this.testFailureModes(failureTask.target, failureTask.options);
+    }
+    /**
+     * Test failure modes in workflow
+     */
+    async testFailureModes(workflowId, options, providedWorkflow) {
+        const workflow = providedWorkflow || await this.getWorkflow(workflowId);
+        const findings = [];
+        const recommendations = [];
+        // 1. Analyze retry configurations
+        const retryAnalysis = this.analyzeRetryConfigurations(workflow);
+        findings.push(...this.retryConfigToFindings(retryAnalysis));
+        // 2. Analyze error branches
+        const errorBranchAnalysis = this.analyzeErrorBranches(workflow);
+        findings.push(...this.errorBranchToFindings(errorBranchAnalysis, workflow));
+        // 3. Analyze continue-on-fail settings
+        const continueOnFailAnalysis = this.analyzeContinueOnFail(workflow);
+        findings.push(...this.continueOnFailToFindings(continueOnFailAnalysis));
+        // 4. Check for partial failure risks in loops
+        if (options?.testPartialFailures !== false) {
+            findings.push(...this.checkPartialFailureRisks(workflow));
+        }
+        // 5. Check for timeout risks
+        if (options?.testTimeouts !== false) {
+            findings.push(...this.checkTimeoutRisks(workflow));
+        }
+        // 6. Check for dead-letter patterns
+        findings.push(...this.checkDeadLetterPatterns(workflow));
+        // 7. Analyze DLQ patterns (enhanced)
+        const dlqAnalysis = this.analyzeDLQPatterns(workflow);
+        // 8. Execute fault injection tests if requested and not dry run
+        let faultInjectionResults;
+        if (options?.injectFaults && !options.dryRun) {
+            faultInjectionResults = await this.executeFaultInjectionTests(workflowId, options.injectFaults, workflow);
+            // Add findings from fault injection results
+            for (const faultResult of faultInjectionResults) {
+                if (!faultResult.errorHandled && faultResult.executed) {
+                    findings.push({
+                        type: 'missing-error-handler',
+                        severity: 'high',
+                        node: faultResult.fault.targetNode,
+                        message: `Fault injection test failed: ${faultResult.fault.faultType} was not handled`,
+                        details: faultResult.details,
+                        suggestion: 'Add error handling (retry, error branch, or continue-on-fail) for this node',
+                    });
+                }
+                if (faultResult.dataIntegrity === 'corrupted') {
+                    findings.push({
+                        type: 'cascade-failure',
+                        severity: 'critical',
+                        node: faultResult.fault.targetNode,
+                        message: `Data corruption detected after ${faultResult.fault.faultType} fault`,
+                        details: `Error propagated to: ${faultResult.errorPropagation.join(' -> ')}`,
+                        suggestion: 'Add data validation and rollback mechanisms',
+                    });
+                }
+            }
+        }
+        // Generate recommendations
+        recommendations.push(...this.generateRecommendations(findings, retryAnalysis, errorBranchAnalysis, continueOnFailAnalysis));
+        recommendations.push(...dlqAnalysis.recommendations);
+        // Calculate scores
+        const errorHandlingScore = this.calculateErrorHandlingScore(errorBranchAnalysis, retryAnalysis);
+        const resilienceScore = this.calculateResilienceScore(findings);
+        const score = Math.round((errorHandlingScore + resilienceScore) / 2);
+        const result = {
+            workflowId: workflow.id || workflowId,
+            workflowName: workflow.name,
+            testDate: new Date().toISOString(),
+            passed: findings.filter(f => f.severity === 'critical' || f.severity === 'high').length === 0,
+            score,
+            errorHandlingScore,
+            resilienceScore,
+            findings,
+            retryAnalysis,
+            errorBranchAnalysis,
+            continueOnFailAnalysis,
+            recommendations,
+            faultInjectionResults,
+            dlqAnalysis,
+        };
+        // Store result
+        await this.storeTestResult(`failure-mode-test:${workflowId}`, result);
+        // Emit event
+        this.emitEvent('failure-mode.test.completed', {
+            workflowId,
+            passed: result.passed,
+            errorHandlingScore,
+            resilienceScore,
+            findingCount: findings.length,
+        });
+        return result;
+    }
+    /**
+     * Analyze retry configurations
+     */
+    analyzeRetryConfigurations(workflow) {
+        const nodesWithRetry = [];
+        const nodesWithoutRetry = [];
+        const retryConfigurations = [];
+        for (const node of workflow.nodes) {
+            const isErrorProne = ERROR_PRONE_NODE_TYPES.some(t => node.type.includes(t));
+            const settings = node.parameters;
+            // Check for retry settings
+            const hasRetry = settings.options && typeof settings.options === 'object' &&
+                ('retry' in settings.options || 'retryOnFail' in settings.options);
+            if (hasRetry) {
+                nodesWithRetry.push(node.name);
+                const retryOptions = (settings.options || {});
+                const retrySubOptions = (retryOptions.retry || {});
+                const issues = [];
+                // Analyze retry configuration
+                const maxRetries = ((retryOptions.maxTries ?? retrySubOptions.maxTries) || 3);
+                const waitBetween = ((retryOptions.waitBetweenTries ?? retrySubOptions.waitBetweenTries) || 1000);
+                if (maxRetries > 10) {
+                    issues.push('Excessive retry count may cause long execution times');
+                }
+                if (maxRetries < 2 && isErrorProne) {
+                    issues.push('Low retry count for error-prone node type');
+                }
+                if (waitBetween < 500) {
+                    issues.push('Short wait between retries may trigger rate limits');
+                }
+                retryConfigurations.push({
+                    node: node.name,
+                    maxRetries,
+                    backoffType: 'fixed', // n8n uses fixed by default
+                    waitBetween,
+                    issues,
+                });
+            }
+            else if (isErrorProne) {
+                nodesWithoutRetry.push(node.name);
+            }
+        }
+        const errorProneNodes = workflow.nodes.filter(n => ERROR_PRONE_NODE_TYPES.some(t => n.type.includes(t))).length;
+        const overallScore = errorProneNodes > 0
+            ? Math.round((nodesWithRetry.length / errorProneNodes) * 100)
+            : 100;
+        return {
+            nodesWithRetry,
+            nodesWithoutRetry,
+            retryConfigurations,
+            overallScore,
+        };
+    }
+    /**
+     * Analyze error branches
+     */
+    analyzeErrorBranches(workflow) {
+        const nodesWithErrorBranch = [];
+        const nodesWithoutErrorBranch = [];
+        const errorHandlers = [];
+        const orphanedErrorBranches = [];
+        for (const node of workflow.nodes) {
+            const isErrorProne = ERROR_PRONE_NODE_TYPES.some(t => node.type.includes(t));
+            // Check if node has error output connection
+            const connections = workflow.connections[node.name];
+            const hasErrorBranch = connections?.main && connections.main.length > 1;
+            if (hasErrorBranch) {
+                nodesWithErrorBranch.push(node.name);
+                // Check if error branch leads somewhere
+                const errorOutputConnections = connections.main[1];
+                if (!errorOutputConnections || errorOutputConnections.length === 0) {
+                    orphanedErrorBranches.push(node.name);
+                }
+                else {
+                    for (const conn of errorOutputConnections) {
+                        errorHandlers.push({
+                            triggerNode: node.name,
+                            handlerNode: conn.node,
+                            handlesAllErrors: true, // Would need runtime analysis to determine
+                        });
+                    }
+                }
+            }
+            else if (isErrorProne) {
+                nodesWithoutErrorBranch.push(node.name);
+            }
+        }
+        const errorProneNodes = workflow.nodes.filter(n => ERROR_PRONE_NODE_TYPES.some(t => n.type.includes(t))).length;
+        const errorBranchCoverage = errorProneNodes > 0
+            ? Math.round((nodesWithErrorBranch.length / errorProneNodes) * 100)
+            : 100;
+        return {
+            nodesWithErrorBranch,
+            nodesWithoutErrorBranch,
+            errorBranchCoverage,
+            orphanedErrorBranches,
+            errorHandlers,
+        };
+    }
+    /**
+     * Analyze continue-on-fail settings
+     */
+    analyzeContinueOnFail(workflow) {
+        const nodesWithContinueOnFail = [];
+        const silentFailureRisks = [];
+        const dataIntegrityRisks = [];
+        for (const node of workflow.nodes) {
+            const settings = node.parameters;
+            const hasContinueOnFail = settings.continueOnFail === true ||
+                (settings.options && settings.options.continueOnFail === true);
+            if (hasContinueOnFail) {
+                nodesWithContinueOnFail.push(node.name);
+                // Check if there's error handling after this node
+                const connections = workflow.connections[node.name];
+                const hasErrorHandling = connections?.main && connections.main.length > 1;
+                if (!hasErrorHandling) {
+                    silentFailureRisks.push(node.name);
+                }
+                // Find downstream nodes that might receive bad data
+                const downstream = this.getDownstreamNodes(workflow, node.name);
+                if (downstream.length > 0) {
+                    dataIntegrityRisks.push({
+                        node: node.name,
+                        risk: 'Continue on fail may pass incomplete/bad data downstream',
+                        affectedDownstream: downstream.map(n => n.name),
+                    });
+                }
+            }
+        }
+        return {
+            nodesWithContinueOnFail,
+            silentFailureRisks,
+            dataIntegrityRisks,
+        };
+    }
+    /**
+     * Check for partial failure risks in loops
+     */
+    checkPartialFailureRisks(workflow) {
+        const findings = [];
+        // Find loop nodes
+        const loopNodes = workflow.nodes.filter(n => LOOP_NODE_TYPES.some(t => n.type.includes(t)));
+        for (const loopNode of loopNodes) {
+            // Check if nodes inside loop have error handling
+            const downstream = this.getDownstreamNodes(workflow, loopNode.name);
+            const errorProneInLoop = downstream.filter(n => ERROR_PRONE_NODE_TYPES.some(t => n.type.includes(t)));
+            for (const errorProneNode of errorProneInLoop) {
+                const connections = workflow.connections[errorProneNode.name];
+                const hasErrorBranch = connections?.main && connections.main.length > 1;
+                if (!hasErrorBranch) {
+                    findings.push({
+                        type: 'partial-failure-risk',
+                        severity: 'high',
+                        node: errorProneNode.name,
+                        message: `Error-prone node in loop without error handling`,
+                        details: `Node "${errorProneNode.name}" is inside loop "${loopNode.name}" but has no error branch. A single item failure could stop the entire loop.`,
+                        suggestion: 'Add error handling or use "continue on fail" with error logging',
+                    });
+                }
+            }
+        }
+        return findings;
+    }
+    /**
+     * Check for timeout risks
+     */
+    checkTimeoutRisks(workflow) {
+        const findings = [];
+        for (const node of workflow.nodes) {
+            // Check HTTP nodes for timeout settings
+            if (node.type.includes('httpRequest')) {
+                const settings = node.parameters;
+                const options = settings.options;
+                const timeout = options?.timeout;
+                if (!timeout) {
+                    findings.push({
+                        type: 'timeout-risk',
+                        severity: 'medium',
+                        node: node.name,
+                        message: 'HTTP request without explicit timeout',
+                        details: 'Missing timeout may cause workflow to hang indefinitely',
+                        suggestion: 'Set explicit timeout in node options',
+                    });
+                }
+                else if (timeout > 300000) { // 5 minutes
+                    findings.push({
+                        type: 'timeout-risk',
+                        severity: 'low',
+                        node: node.name,
+                        message: 'HTTP request with very long timeout',
+                        details: `Timeout set to ${timeout / 1000}s which may cause long hangs`,
+                        suggestion: 'Consider shorter timeout with retry logic',
+                    });
+                }
+            }
+            // Check execute command nodes
+            if (node.type.includes('executeCommand')) {
+                findings.push({
+                    type: 'timeout-risk',
+                    severity: 'medium',
+                    node: node.name,
+                    message: 'Execute command without timeout control',
+                    details: 'Shell commands can hang indefinitely',
+                    suggestion: 'Add timeout to command or use n8n timeout settings',
+                });
+            }
+        }
+        return findings;
+    }
+    /**
+     * Check for dead-letter patterns
+     */
+    checkDeadLetterPatterns(workflow) {
+        const findings = [];
+        // Check if workflow has any error notification/logging at the end
+        const hasErrorWorkflow = workflow.settings?.errorWorkflow;
+        const hasErrorNotification = workflow.nodes.some(n => n.type.includes('email') || n.type.includes('slack') || n.type.includes('telegram'));
+        // Check for nodes that send to DLQ or error queue
+        const hasDLQPattern = workflow.nodes.some(n => {
+            const params = JSON.stringify(n.parameters).toLowerCase();
+            return params.includes('dlq') || params.includes('dead') || params.includes('error-queue');
+        });
+        if (!hasErrorWorkflow && !hasErrorNotification && !hasDLQPattern) {
+            findings.push({
+                type: 'dlq-missing',
+                severity: 'medium',
+                node: 'workflow',
+                message: 'No dead-letter or error notification pattern detected',
+                details: 'Failed executions may go unnoticed without error notifications',
+                suggestion: 'Add error workflow in workflow settings or add notification node for failures',
+            });
+        }
+        return findings;
+    }
+    /**
+     * Execute fault injection tests
+     */
+    async executeFaultInjection(workflow, faults) {
+        const findings = [];
+        // Note: Actual fault injection would require modifying workflow or using n8n's testing features
+        // This is a placeholder for the concept
+        for (const fault of faults) {
+            const targetNode = workflow.nodes.find(n => n.name === fault.targetNode);
+            if (!targetNode) {
+                findings.push({
+                    type: 'missing-error-handler',
+                    severity: 'low',
+                    node: fault.targetNode,
+                    message: `Fault injection target "${fault.targetNode}" not found`,
+                    details: 'Cannot inject fault to non-existent node',
+                    suggestion: 'Check node name spelling',
+                });
+                continue;
+            }
+            // Check if node has error handling for the injected fault type
+            const connections = workflow.connections[fault.targetNode];
+            const hasErrorBranch = connections?.main && connections.main.length > 1;
+            if (!hasErrorBranch) {
+                findings.push({
+                    type: 'missing-error-handler',
+                    severity: 'high',
+                    node: fault.targetNode,
+                    message: `No error handler for ${fault.faultType} fault`,
+                    details: `If ${fault.faultType} occurs, the workflow will fail without graceful handling`,
+                    suggestion: 'Add error branch to handle this failure mode',
+                });
+            }
+        }
+        return findings;
+    }
+    /**
+     * Convert retry analysis to findings
+     */
+    retryConfigToFindings(analysis) {
+        const findings = [];
+        for (const node of analysis.nodesWithoutRetry) {
+            findings.push({
+                type: 'retry-misconfiguration',
+                severity: 'medium',
+                node,
+                message: 'Error-prone node without retry configuration',
+                details: 'Transient failures will cause immediate workflow failure',
+                suggestion: 'Add retry settings in node options',
+            });
+        }
+        for (const config of analysis.retryConfigurations) {
+            for (const issue of config.issues) {
+                findings.push({
+                    type: 'retry-misconfiguration',
+                    severity: 'low',
+                    node: config.node,
+                    message: issue,
+                    details: `Current retry config: ${config.maxRetries} retries, ${config.waitBetween}ms wait`,
+                    suggestion: 'Adjust retry settings based on expected failure modes',
+                });
+            }
+        }
+        return findings;
+    }
+    /**
+     * Convert error branch analysis to findings
+     */
+    errorBranchToFindings(analysis, workflow) {
+        const findings = [];
+        for (const node of analysis.nodesWithoutErrorBranch) {
+            findings.push({
+                type: 'missing-error-handler',
+                severity: 'medium',
+                node,
+                message: 'Error-prone node without error branch',
+                details: 'Errors will propagate to workflow level without graceful handling',
+                suggestion: 'Add error branch to handle failures gracefully',
+            });
+        }
+        for (const orphan of analysis.orphanedErrorBranches) {
+            findings.push({
+                type: 'missing-error-handler',
+                severity: 'low',
+                node: orphan,
+                message: 'Error branch exists but is not connected',
+                details: 'Error output has no target node',
+                suggestion: 'Connect error branch to an error handler node',
+            });
+        }
+        return findings;
+    }
+    /**
+     * Convert continue-on-fail analysis to findings
+     */
+    continueOnFailToFindings(analysis) {
+        const findings = [];
+        for (const node of analysis.silentFailureRisks) {
+            findings.push({
+                type: 'silent-failure',
+                severity: 'high',
+                node,
+                message: 'Continue on fail without error tracking',
+                details: 'Failures will be silently ignored without logging or notification',
+                suggestion: 'Add error logging/notification when using continue on fail',
+            });
+        }
+        for (const risk of analysis.dataIntegrityRisks) {
+            findings.push({
+                type: 'cascade-failure',
+                severity: 'medium',
+                node: risk.node,
+                message: risk.risk,
+                details: `May affect: ${risk.affectedDownstream.join(', ')}`,
+                suggestion: 'Add data validation after nodes with continue on fail',
+            });
+        }
+        return findings;
+    }
+    /**
+     * Calculate error handling score
+     */
+    calculateErrorHandlingScore(errorBranchAnalysis, retryAnalysis) {
+        const errorBranchScore = errorBranchAnalysis.errorBranchCoverage;
+        const retryScore = retryAnalysis.overallScore;
+        return Math.round((errorBranchScore + retryScore) / 2);
+    }
+    /**
+     * Calculate resilience score
+     */
+    calculateResilienceScore(findings) {
+        const criticalCount = findings.filter(f => f.severity === 'critical').length;
+        const highCount = findings.filter(f => f.severity === 'high').length;
+        const mediumCount = findings.filter(f => f.severity === 'medium').length;
+        return Math.max(0, 100 - (criticalCount * 25) - (highCount * 15) - (mediumCount * 5));
+    }
+    /**
+     * Generate recommendations
+     */
+    generateRecommendations(findings, retryAnalysis, errorBranchAnalysis, continueOnFailAnalysis) {
+        const recommendations = [];
+        if (retryAnalysis.nodesWithoutRetry.length > 0) {
+            recommendations.push(`Add retry configuration to ${retryAnalysis.nodesWithoutRetry.length} error-prone nodes`);
+        }
+        if (errorBranchAnalysis.errorBranchCoverage < 50) {
+            recommendations.push('Error branch coverage is low - add error handling to critical nodes');
+        }
+        if (continueOnFailAnalysis.silentFailureRisks.length > 0) {
+            recommendations.push('Add error logging when using "continue on fail" to avoid silent failures');
+        }
+        const dlqFindings = findings.filter(f => f.type === 'dlq-missing');
+        if (dlqFindings.length > 0) {
+            recommendations.push('Consider adding error workflow or notification for failed executions');
+        }
+        return recommendations;
+    }
+    // ============================================================================
+    // Active Fault Injection Testing
+    // ============================================================================
+    /**
+     * Execute fault injection tests against a live workflow
+     * This actually runs the workflow with injected faults to verify error handling
+     */
+    async executeFaultInjectionTests(workflowId, faults, providedWorkflow) {
+        const results = [];
+        const workflow = providedWorkflow || await this.getWorkflow(workflowId);
+        // If no faults specified, auto-generate faults for error-prone nodes
+        const faultsToTest = faults || this.generateDefaultFaults(workflow);
+        if (faultsToTest.length === 0) {
+            return results;
+        }
+        // Create test harness
+        const harness = new N8nTestHarness_1.N8nTestHarness(this.n8nConfig);
+        try {
+            for (const fault of faultsToTest) {
+                const result = await this.executeSingleFaultTest(harness, workflowId, fault, workflow);
+                results.push(result);
+            }
+        }
+        finally {
+            await harness.cleanup();
+        }
+        return results;
+    }
+    /**
+     * Generate default faults for error-prone nodes
+     */
+    generateDefaultFaults(workflow) {
+        const faults = [];
+        for (const node of workflow.nodes) {
+            const isErrorProne = ERROR_PRONE_NODE_TYPES.some(t => node.type.includes(t));
+            if (!isErrorProne)
+                continue;
+            // Add error fault
+            faults.push({
+                targetNode: node.name,
+                faultType: 'error',
+                errorMessage: `Simulated error in ${node.name}`,
+            });
+            // Add timeout fault for HTTP nodes
+            if (node.type.includes('httpRequest')) {
+                faults.push({
+                    targetNode: node.name,
+                    faultType: 'timeout',
+                    delay: 5000,
+                });
+            }
+            // Add empty response fault
+            faults.push({
+                targetNode: node.name,
+                faultType: 'empty-response',
+            });
+        }
+        return faults;
+    }
+    /**
+     * Execute a single fault injection test
+     */
+    async executeSingleFaultTest(harness, workflowId, fault, workflow) {
+        const startTime = Date.now();
+        // Convert to harness format
+        const faultConfig = {
+            targetNode: fault.targetNode,
+            faultType: fault.faultType === 'rate-limit' ? 'rate-limit' : fault.faultType,
+            probability: fault.probability,
+            delay: fault.delay,
+            errorMessage: fault.errorMessage,
+        };
+        try {
+            const testResult = await harness.executeWithFaults(workflowId, [faultConfig]);
+            // Analyze the result
+            return this.analyzeFaultTestResult(fault, testResult, workflow, Date.now() - startTime);
+        }
+        catch (error) {
+            return {
+                fault,
+                executed: false,
+                errorHandled: false,
+                retryAttempts: 0,
+                finalStatus: 'crashed',
+                executionTime: Date.now() - startTime,
+                errorPropagation: [],
+                dataIntegrity: 'corrupted',
+                details: `Failed to execute fault test: ${error.message}`,
+            };
+        }
+    }
+    /**
+     * Analyze the result of a fault injection test
+     */
+    analyzeFaultTestResult(fault, testResult, workflow, executionTime) {
+        const execution = testResult.execution;
+        if (!execution) {
+            return {
+                fault,
+                executed: true,
+                errorHandled: false,
+                retryAttempts: 0,
+                finalStatus: testResult.error?.includes('timeout') ? 'timeout' : 'crashed',
+                executionTime,
+                errorPropagation: [],
+                dataIntegrity: 'corrupted',
+                details: testResult.error || 'Execution failed with no result',
+            };
+        }
+        // Analyze execution data
+        const runData = execution.data?.resultData?.runData || {};
+        const errorNodes = this.findErrorNodes(runData);
+        const retryAttempts = this.countRetryAttempts(runData, fault.targetNode);
+        // Check if error was handled
+        const hasErrorBranch = this.checkErrorBranchUsed(workflow, fault.targetNode, runData);
+        const hasContinueOnFail = this.checkContinueOnFailUsed(workflow, fault.targetNode, runData);
+        const errorHandled = hasErrorBranch || hasContinueOnFail || retryAttempts > 0;
+        // Determine final status
+        let finalStatus = 'crashed';
+        if (execution.finished && !execution.data?.resultData?.error) {
+            finalStatus = errorHandled ? 'recovered' : 'failed-gracefully';
+        }
+        else if (errorHandled) {
+            finalStatus = 'failed-gracefully';
+        }
+        // Check data integrity
+        const dataIntegrity = this.assessDataIntegrity(workflow, runData, fault.targetNode);
+        // Track error propagation
+        const errorPropagation = this.traceErrorPropagation(workflow, fault.targetNode, runData);
+        return {
+            fault,
+            executed: true,
+            errorHandled,
+            retryAttempts,
+            finalStatus,
+            executionTime,
+            errorPropagation,
+            dataIntegrity,
+            details: this.generateFaultTestDetails(fault, errorHandled, retryAttempts, finalStatus),
+        };
+    }
+    /**
+     * Find nodes that had errors in execution
+     */
+    findErrorNodes(runData) {
+        const errorNodes = [];
+        for (const [nodeName, runs] of Object.entries(runData)) {
+            const nodeRuns = runs;
+            if (nodeRuns.some(run => run.error)) {
+                errorNodes.push(nodeName);
+            }
+        }
+        return errorNodes;
+    }
+    /**
+     * Count retry attempts for a node
+     */
+    countRetryAttempts(runData, nodeName) {
+        const nodeRuns = runData[nodeName];
+        if (!nodeRuns)
+            return 0;
+        // Multiple runs of the same node indicate retries
+        return Math.max(0, nodeRuns.length - 1);
+    }
+    /**
+     * Check if error branch was used
+     */
+    checkErrorBranchUsed(workflow, faultNode, runData) {
+        const connections = workflow.connections[faultNode];
+        if (!connections?.main || connections.main.length < 2)
+            return false;
+        // Check if error output (index 1) was executed
+        const errorOutputConnections = connections.main[1];
+        if (!errorOutputConnections)
+            return false;
+        return errorOutputConnections.some(conn => {
+            const connectedNodeRuns = runData[conn.node];
+            return connectedNodeRuns && connectedNodeRuns.length > 0;
+        });
+    }
+    /**
+     * Check if continue-on-fail was used
+     */
+    checkContinueOnFailUsed(workflow, faultNode, runData) {
+        const node = workflow.nodes.find(n => n.name === faultNode);
+        if (!node)
+            return false;
+        const params = node.parameters;
+        const hasContinueOnFail = params.continueOnFail === true ||
+            (params.options && params.options.continueOnFail === true);
+        if (!hasContinueOnFail)
+            return false;
+        // Check if downstream nodes executed despite error
+        const connections = workflow.connections[faultNode];
+        if (!connections?.main?.[0])
+            return false;
+        return connections.main[0].some(conn => {
+            const connectedNodeRuns = runData[conn.node];
+            return connectedNodeRuns && connectedNodeRuns.length > 0;
+        });
+    }
+    /**
+     * Assess data integrity after fault
+     */
+    assessDataIntegrity(workflow, runData, faultNode) {
+        // Get downstream nodes
+        const downstream = this.getDownstreamNodes(workflow, faultNode);
+        // Check if any downstream nodes ran
+        const downstreamRan = downstream.filter(node => {
+            const runs = runData[node.name];
+            return runs && runs.length > 0;
+        });
+        if (downstreamRan.length === 0) {
+            // Error stopped propagation - data preserved (nothing bad happened downstream)
+            return 'preserved';
+        }
+        // Check if downstream received valid data
+        let hasValidData = true;
+        let hasCorruptedData = false;
+        for (const node of downstreamRan) {
+            const runs = runData[node.name];
+            for (const run of runs) {
+                if (run.error) {
+                    hasCorruptedData = true;
+                }
+                const outputData = run.data?.main?.[0]?.[0]?.json;
+                if (outputData === undefined || outputData === null) {
+                    hasValidData = false;
+                }
+            }
+        }
+        if (hasCorruptedData)
+            return 'corrupted';
+        if (!hasValidData)
+            return 'partial-loss';
+        return 'preserved';
+    }
+    /**
+     * Trace how error propagated through workflow
+     */
+    traceErrorPropagation(workflow, faultNode, runData) {
+        const propagation = [faultNode];
+        const visited = new Set([faultNode]);
+        const queue = [faultNode];
+        while (queue.length > 0) {
+            const current = queue.shift();
+            const connections = workflow.connections[current];
+            if (!connections?.main)
+                continue;
+            for (const outputs of connections.main) {
+                if (!outputs)
+                    continue;
+                for (const conn of outputs) {
+                    if (visited.has(conn.node))
+                        continue;
+                    visited.add(conn.node);
+                    // Check if this node was affected by the error
+                    const runs = runData[conn.node];
+                    if (runs?.some(r => r.error)) {
+                        propagation.push(conn.node);
+                        queue.push(conn.node);
+                    }
+                }
+            }
+        }
+        return propagation;
+    }
+    /**
+     * Generate human-readable details for fault test
+     */
+    generateFaultTestDetails(fault, errorHandled, retryAttempts, finalStatus) {
+        const parts = [];
+        parts.push(`Injected ${fault.faultType} fault into "${fault.targetNode}".`);
+        if (retryAttempts > 0) {
+            parts.push(`Node retried ${retryAttempts} time(s).`);
+        }
+        if (errorHandled) {
+            parts.push('Error was handled by workflow.');
+        }
+        else {
+            parts.push('Error was NOT handled - no error branch or retry configured.');
+        }
+        parts.push(`Final status: ${finalStatus}.`);
+        return parts.join(' ');
+    }
+    /**
+     * Analyze Dead Letter Queue patterns
+     */
+    analyzeDLQPatterns(workflow) {
+        const dlqNodes = [];
+        const unhandledFailurePaths = [];
+        const recommendations = [];
+        // Look for DLQ-like patterns
+        for (const node of workflow.nodes) {
+            const nameLower = node.name.toLowerCase();
+            const typeLower = node.type.toLowerCase();
+            // Common DLQ indicators
+            if (nameLower.includes('dlq') ||
+                nameLower.includes('dead') ||
+                nameLower.includes('failed') ||
+                nameLower.includes('error-queue') ||
+                nameLower.includes('retry-queue')) {
+                dlqNodes.push(node.name);
+            }
+            // Check for error notification patterns
+            if ((typeLower.includes('slack') || typeLower.includes('email') || typeLower.includes('telegram')) &&
+                (nameLower.includes('error') || nameLower.includes('alert') || nameLower.includes('notify'))) {
+                dlqNodes.push(node.name);
+            }
+        }
+        // Find error-prone nodes without error handling leading to DLQ
+        for (const node of workflow.nodes) {
+            const isErrorProne = ERROR_PRONE_NODE_TYPES.some(t => node.type.includes(t));
+            if (!isErrorProne)
+                continue;
+            const connections = workflow.connections[node.name];
+            const hasErrorBranch = connections?.main && connections.main.length > 1;
+            if (!hasErrorBranch) {
+                unhandledFailurePaths.push(node.name);
+            }
+            else {
+                // Check if error branch leads to DLQ
+                const errorOutputs = connections?.main?.[1] || [];
+                const leadsToDLQ = errorOutputs.some(conn => dlqNodes.includes(conn.node) ||
+                    this.eventuallyLeadsTo(workflow, conn.node, dlqNodes));
+                if (!leadsToDLQ) {
+                    unhandledFailurePaths.push(node.name);
+                }
+            }
+        }
+        // Determine poison message handling
+        let poisonMessageHandling = 'missing';
+        if (dlqNodes.length > 0) {
+            poisonMessageHandling = unhandledFailurePaths.length === 0 ? 'present' : 'partial';
+        }
+        // Generate recommendations
+        if (dlqNodes.length === 0) {
+            recommendations.push('Add a dead-letter queue pattern for failed message handling');
+            recommendations.push('Consider adding error notification (Slack, Email) for critical failures');
+        }
+        if (unhandledFailurePaths.length > 0) {
+            recommendations.push(`Add error branches for: ${unhandledFailurePaths.slice(0, 3).join(', ')}${unhandledFailurePaths.length > 3 ? '...' : ''}`);
+        }
+        return {
+            hasDLQPattern: dlqNodes.length > 0,
+            dlqNodes,
+            unhandledFailurePaths,
+            poisonMessageHandling,
+            recommendations,
+        };
+    }
+    /**
+     * Check if a node eventually leads to any of the target nodes
+     */
+    eventuallyLeadsTo(workflow, startNode, targetNodes) {
+        const visited = new Set();
+        const queue = [startNode];
+        while (queue.length > 0) {
+            const current = queue.shift();
+            if (visited.has(current))
+                continue;
+            visited.add(current);
+            if (targetNodes.includes(current)) {
+                return true;
+            }
+            const connections = workflow.connections[current];
+            if (!connections?.main)
+                continue;
+            for (const outputs of connections.main) {
+                if (!outputs)
+                    continue;
+                for (const conn of outputs) {
+                    queue.push(conn.node);
+                }
+            }
+        }
+        return false;
+    }
+    /**
+     * Quick failure mode check
+     */
+    async quickCheck(workflowId) {
+        const result = await this.testFailureModes(workflowId, { dryRun: true });
+        return {
+            resilient: result.passed,
+            errorHandlingScore: result.errorHandlingScore,
+            criticalIssues: result.findings.filter(f => f.severity === 'critical').length,
+            topIssue: result.findings[0]?.message || null,
+        };
+    }
+}
+exports.N8nFailureModeTesterAgent = N8nFailureModeTesterAgent;
+//# sourceMappingURL=N8nFailureModeTesterAgent.js.map