{ "nodes": [ { "parameters": { "rule": { "interval": [ { "triggerAtHour": 7 } ] } }, "type": "n8n-nodes-base.scheduleTrigger", "typeVersion": 1.2, "position": [ -944, 176 ], "id": "2ff737d7-b4b2-47c7-97db-232f27c51ee2", "name": "Weekly Monday 7AM" }, { "parameters": { "path": "root-cause-mining", "responseMode": "responseNode", "options": {} }, "type": "n8n-nodes-base.webhook", "typeVersion": 2.1, "position": [ -944, 352 ], "id": "9bd8da7b-d5f5-4d62-917b-3ba30ec1607d", "name": "Manual Trigger Webhook" }, { "parameters": { "promptType": "define", "text": "=Today is {{ $now.format('yyyy-MM-dd') }}. Analyze the past 7 days of trade operations data.\n\nSTEP-BY-STEP INSTRUCTIONS:\n1. First, call get_sla_config to know the current SLA thresholds.\n2. Call get_all_lcs with fromDate={{ $now.minus(7, 'days').format('yyyy-MM-dd') }} and toDate={{ $now.format('yyyy-MM-dd') }} to get all L/Cs from the past week.\n3. From the results, identify ALL L/Cs with status 'Breached' or 'Breached with Exception'.\n4. For EACH breached L/C, call get_lc_exceptions with the L/C id to get exception history and root cause details.\n5. Call get_all_events with fromDate={{ $now.minus(7, 'days').format('yyyy-MM-dd') }} and toDate={{ $now.format('yyyy-MM-dd') }} to get the full event timeline.\n6. Call get_staff_workload to understand current staff distribution.\n7. Perform your pattern analysis and produce the final report.\n\nYour FINAL OUTPUT must follow the exact format specified in your system instructions.", "options": { "systemMessage": "You are a Senior Trade Operations Intelligence Analyst specializing in root cause analysis and process mining.\n\nYour mission: Analyze 7 days of L/C processing data to discover SYSTEMIC PATTERNS causing SLA breaches \u2014 not just list individual incidents.\n\n## ANALYSIS FRAMEWORK\n\nPerform these analyses using the data you collect:\n\n### 1. BOTTLENECK STAGE ANALYSIS\n- Calculate average dwell time per stage: Received\u2192Drafting, Drafting\u2192Checking Underlying, Checking\u2192Released\n- Use the timestamps: receivedAt, draftingStartedAt, checkingStartedAt, releasedAt\n- Identify which stage transition consistently takes the longest\n- Flag stages where dwell time exceeds 50% of total SLA budget\n\n### 2. EXCEPTION PATTERN MINING\n- Group all exceptions by their 'reason' field from lc_exceptions\n- Calculate: frequency count, average resolution time (resolutionMinutes), repeat offender L/Cs\n- Identify the TOP 3 most common exception reasons\n- Check if certain exception types cluster at specific times of day\n\n### 3. STAFF WORKLOAD CORRELATION\n- Cross-reference breached L/Cs with their assignedTo field\n- Calculate breach rate per staff member: (breached / total assigned) \u00d7 100\n- Identify if breaches correlate with staff who have too many concurrent assignments\n- Check if certain staff-to-transaction-type pairings perform worse\n\n### 4. TEMPORAL PATTERN DETECTION\n- Group breaches by day-of-week and hour-of-day (from receivedAt)\n- Identify if certain receiving times correlate with higher breach rates\n- Check if breaches cluster around specific days (e.g., Monday morning backlog)\n\n### 5. TRANSACTION TYPE BREAKDOWN\n- Compare SLA compliance rates: Import vs Export vs Bank Guarantee\n- Identify if one transaction type has disproportionately more breaches\n- Cross-reference with the different SLA thresholds per type\n\n## OUTPUT FORMAT (STRICT)\n\nYour output MUST follow this exact structure:\n\n---\n\n# \ud83d\udcca Weekly Root Cause Analysis Report\n**Period:** [fromDate] \u2014 [toDate]\n**Generated:** [current datetime]\n\n## \ud83d\udcc8 Executive Summary\n[3 sentences: total L/Cs processed, breach count & rate, the single biggest systemic issue found]\n\n## \ud83d\udd34 Key Findings\n\n### 1. Primary Bottleneck\n- **Stage:** [stage name]\n- **Average Dwell Time:** [X minutes]\n- **Impact:** [how many breaches this caused]\n- **Evidence:** [specific URNs and data points]\n\n### 2. Top Exception Patterns\n| Rank | Exception Reason | Count | Avg Resolution (min) | Affected Staff |\n|------|-----------------|-------|---------------------|----------------|\n| 1 | ... | ... | ... | ... |\n| 2 | ... | ... | ... | ... |\n| 3 | ... | ... | ... | ... |\n\n### 3. Staff Performance Matrix\n| Staff | Total Assigned | Breached | Breach Rate | Avg Processing Time |\n|-------|---------------|----------|-------------|--------------------|\n| ... | ... | ... | ... | ... |\n\n### 4. Temporal Hotspots\n- **Worst Day:** [day] ([X]% of weekly breaches)\n- **Worst Hour Window:** [HH:00 - HH:00]\n- **Pattern:** [description]\n\n### 5. Transaction Type Health\n| Type | Total | Breached | Compliance Rate | SLA Threshold |\n|------|-------|----------|----------------|---------------|\n| Import | ... | ... | ... | ... min |\n| Export | ... | ... | ... | ... min |\n| Bank Guarantee | ... | ... | ... | ... min |\n\n## \ud83d\udca1 Strategic Recommendations\n1. **[Action 1]:** [Specific, measurable recommendation with expected impact]\n2. **[Action 2]:** [Specific, measurable recommendation with expected impact]\n3. **[Action 3]:** [Specific, measurable recommendation with expected impact]\n\n## \ud83d\udd04 Trend vs Previous Period\n[If data allows, note whether breach rate is improving or worsening]\n\n---\n\n# \ud83d\udcca Laporan Analisis Akar Penyebab Mingguan\n[Full Bahasa Indonesia translation of the above report with identical structure]\n\n---\n\nCRITICAL RULES:\n- DO NOT output your internal reasoning, chain-of-thought, or data dumps\n- DO NOT include raw JSON responses in your output\n- ONLY output the formatted report above\n- Use REAL data from the API calls, never fabricate numbers\n- If data is insufficient for an analysis section, state 'Insufficient data' rather than guessing" } }, "type": "@n8n/n8n-nodes-langchain.agent", "typeVersion": 3.1, "position": [ -432, 176 ], "id": "a0e205dd-1d08-435f-9a57-6daef3a8f9e0", "name": "Root Cause Mining Agent" }, { "parameters": { "modelName": "models/gemini-2.5-pro", "options": { "temperature": 0.2 } }, "type": "@n8n/n8n-nodes-langchain.lmChatGoogleGemini", "typeVersion": 1, "position": [ -608, 416 ], "id": "aff492d0-b988-457d-88e7-b50868c0a05a", "name": "Google Gemini Chat Model", "credentials": { "googlePalmApi": { "name": "" } } }, { "parameters": { "model": "qwen3.5:397b", "options": {} }, "type": "@n8n/n8n-nodes-langchain.lmChatOllama", "typeVersion": 1, "position": [ -496, 416 ], "id": "495a80e5-6122-44d7-ac67-9cb6914aaa9a", "name": "Ollama Chat Model", "credentials": { "ollamaApi": { "name": "" } } }, { "parameters": { "toolDescription": "Fetches all L/Cs within a date range. Supports query params: status, transactionType, limit, offset, fromDate, toDate. Use limit=500 to get all records. Returns array of L/C objects with fields: id, urn, status, assignedTo, transactionType, receivedAt, draftingStartedAt, checkingStartedAt, releasedAt, exceptionTotalMinutes, exceptionReason, approvedBy.", "url": "=http://localhost:8081/api/lc?limit=500&fromDate={{$fromAI(\"fromDate\",\"Start date in YYYY-MM-DD format\")}}&toDate={{$fromAI(\"toDate\",\"End date in YYYY-MM-DD format\")}}", "options": {} }, "type": "n8n-nodes-base.httpRequestTool", "typeVersion": 4.4, "position": [ -352, 416 ], "id": "ef803214-1389-428a-82be-f795dd965235", "name": "get_all_lcs" }, { "parameters": { "toolDescription": "Fetches all event logs within a date range. Events track every status change for every L/C. Each event has: id, lcId, urn, user, action, from (previous status), to (new status), notes, timestamp. Use limit=500 and provide fromDate and toDate to get a full week of events.", "url": "=http://localhost:8081/api/events?limit=500&fromDate={{$fromAI(\"fromDate\",\"Start date in YYYY-MM-DD format\")}}&toDate={{$fromAI(\"toDate\",\"End date in YYYY-MM-DD format\")}}", "options": {} }, "type": "n8n-nodes-base.httpRequestTool", "typeVersion": 4.4, "position": [ -208, 416 ], "id": "e7f0caa5-b208-4e4e-b031-7cc8f96106ef", "name": "get_all_events" }, { "parameters": { "toolDescription": "Fetches the exception history for a specific L/C by its numeric ID. Returns an array of exception objects with: id, lcId, reason, startedAt, resolvedAt, resolutionMinutes, resolvedToStatus, resolvedBy. Use this to investigate WHY a specific L/C was delayed. Requires the L/C numeric ID (not URN).", "url": "=http://localhost:8081/api/lc/{{$fromAI(\"lcId\",\"The numeric ID of the L/C\")}}/exceptions", "options": {} }, "type": "n8n-nodes-base.httpRequestTool", "typeVersion": 4.4, "position": [ -64, 416 ], "id": "1e2205a4-94d7-4427-9acd-44f9a2ea90f5", "name": "get_lc_exceptions" }, { "parameters": { "toolDescription": "Fetches the list of trade operations staff (assignees) and their section. Returns array with fields: id, name, section, isActive. Use this to understand team structure and cross-reference with breach data to find overloaded staff.", "url": "http://localhost:8081/api/assignees", "options": {} }, "type": "n8n-nodes-base.httpRequestTool", "typeVersion": 4.4, "position": [ 80, 416 ], "id": "d5b5736e-105a-4f49-88cd-30cfd36934ab", "name": "get_staff_workload" }, { "parameters": { "toolDescription": "Fetches the current SLA configuration. Returns: importSlaMaxMinutes (default 120), exportSlaMaxMinutes, bgSlaMaxMinutes. Use this to know the threshold for each transaction type.", "url": "http://localhost:8081/api/sla", "options": {} }, "type": "n8n-nodes-base.httpRequestTool", "typeVersion": 4.4, "position": [ 224, 416 ], "id": "2597a3fd-f061-4e84-a72a-9c7c4e44b200", "name": "get_sla_config" }, { "parameters": { "assignments": { "assignments": [ { "id": "report", "name": "report", "value": "={{ $json.output }}", "type": "string" }, { "id": "generatedAt", "name": "generatedAt", "value": "={{ $now.format('yyyy-MM-dd HH:mm:ss') }}", "type": "string" } ] }, "options": {} }, "type": "n8n-nodes-base.set", "typeVersion": 3.4, "position": [ -32, 176 ], "id": "9923de2a-8705-48cd-8ab6-f3e04b0461c5", "name": "Format Output" }, { "parameters": { "respondWith": "text", "responseBody": "={{ $json.report }}", "options": {} }, "type": "n8n-nodes-base.respondToWebhook", "typeVersion": 1.5, "position": [ 304, 240 ], "id": "ea45c946-690c-43ed-909c-cf676954a158", "name": "Respond to Webhook" }, { "parameters": { "subject": "=\ud83d\udcca Weekly Root Cause Analysis Report - {{ $now.format('yyyy-MM-dd') }}", "options": {} }, "type": "n8n-nodes-base.emailSend", "typeVersion": 2.1, "position": [ 320, 64 ], "id": "3e7baa9f-f1e6-450b-b9aa-c21a1926e38f", "name": "Email Report", "credentials": { "smtp": { "name": "" } }, "disabled": true } ], "connections": { "Weekly Monday 7AM": { "main": [ [ { "node": "Root Cause Mining Agent", "type": "main", "index": 0 } ] ] }, "Manual Trigger Webhook": { "main": [ [ { "node": "Root Cause Mining Agent", "type": "main", "index": 0 } ] ] }, "Root Cause Mining Agent": { "main": [ [ { "node": "Format Output", "type": "main", "index": 0 } ] ] }, "Google Gemini Chat Model": { "ai_languageModel": [ [ { "node": "Root Cause Mining Agent", "type": "ai_languageModel", "index": 0 } ] ] }, "get_all_lcs": { "ai_tool": [ [ { "node": "Root Cause Mining Agent", "type": "ai_tool", "index": 0 } ] ] }, "get_all_events": { "ai_tool": [ [ { "node": "Root Cause Mining Agent", "type": "ai_tool", "index": 0 } ] ] }, "get_lc_exceptions": { "ai_tool": [ [ { "node": "Root Cause Mining Agent", "type": "ai_tool", "index": 0 } ] ] }, "get_staff_workload": { "ai_tool": [ [ { "node": "Root Cause Mining Agent", "type": "ai_tool", "index": 0 } ] ] }, "get_sla_config": { "ai_tool": [ [ { "node": "Root Cause Mining Agent", "type": "ai_tool", "index": 0 } ] ] }, "Format Output": { "main": [ [ { "node": "Email Report", "type": "main", "index": 0 }, { "node": "Respond to Webhook", "type": "main", "index": 0 } ] ] } }, "meta": { "templateCredsSetupCompleted": true } }