{ "edges": [ { "to": "85a3602f-4211-42cd-b0c0-7952e8ec3d1d", "from": "93828410-1aa3-4d06-9b45-e7a4537b5f3b", "toPort": "0", "fromPort": "0" }, { "to": "a20b8b36-a5fe-454c-9df9-a08faf10dba8", "from": "85a3602f-4211-42cd-b0c0-7952e8ec3d1d", "toPort": "0", "fromPort": "0" }, { "to": "1b7f0641-f611-4321-ba25-d93c6f46b844", "from": "a20b8b36-a5fe-454c-9df9-a08faf10dba8", "toPort": "0", "fromPort": "0" }, { "to": "a20b8b36-a5fe-454c-9df9-a08faf10dba8", "from": "ca58e161-67a4-4c9c-a60f-228d6da4698c", "toPort": "0", "fromPort": "ai_tool" }, { "to": "93828410-1aa3-4d06-9b45-e7a4537b5f3b", "from": "a2406fd5-fb89-46e1-a7f2-db40396945a7", "toPort": "1", "fromPort": "0" }, { "to": "a20b8b36-a5fe-454c-9df9-a08faf10dba8", "from": "e9d25b17-61f6-4be2-968b-7a5607026855", "toPort": "0", "fromPort": "ai_tool" }, { "to": "62d40c47-6da6-440e-9621-1e9e853e058d", "from": "da526997-7792-4d84-a0b2-b4cb8d9b2ef9", "toPort": "0", "fromPort": "0" }, { "to": "da526997-7792-4d84-a0b2-b4cb8d9b2ef9", "from": "1b7f0641-f611-4321-ba25-d93c6f46b844", "toPort": "0", "fromPort": "0" }, { "to": "a20b8b36-a5fe-454c-9df9-a08faf10dba8", "from": "473df4d2-a9ba-4bb0-81ff-1da25d3e46e1", "toPort": "0", "fromPort": "ai_tool" }, { "to": "4b349210-7f35-4dc0-905f-081d945bb8a8", "from": "780ddba7-3663-425a-9b2a-bd066aa0a5a3", "toPort": "0", "fromPort": "0" }, { "to": "a20b8b36-a5fe-454c-9df9-a08faf10dba8", "from": "271b834e-0881-4016-9002-6529df935d6f", "toPort": "0", "fromPort": "ai_languageModel" }, { "to": "473df4d2-a9ba-4bb0-81ff-1da25d3e46e1", "from": "db631503-b383-4299-a149-28d75642f91f", "toPort": "0", "fromPort": "ai_embedding" }, { "to": "93828410-1aa3-4d06-9b45-e7a4537b5f3b", "from": "4b349210-7f35-4dc0-905f-081d945bb8a8", "toPort": "0", "fromPort": "0" }, { "to": "a2406fd5-fb89-46e1-a7f2-db40396945a7", "from": "4b349210-7f35-4dc0-905f-081d945bb8a8", "toPort": "0", "fromPort": "1" }, { "to": "780ddba7-3663-425a-9b2a-bd066aa0a5a3", "from": "8eac83c2-6b95-4ed9-aae0-39d10d47c2c7", "toPort": "0", "fromPort": "0" } ], "nodes": [ { "id": "8eac83c2-6b95-4ed9-aae0-39d10d47c2c7", "name": "When Executed by Another Workflow", "type": "trigger", "params": { "kind": "n8n-nodes-base.executeWorkflowTrigger", "inputSource": "jsonExample", "jsonExample": "{\n \"message\": \"...\",\n \"post_id\": \"...\",\n \"channel_id\": \"...\",\n \"user_name\": \"...\",\n \"is_thread\": false,\n \"thread_root_id\": \"...\",\n \"category\": \"create_resource\",\n \"confidence\": 0.95,\n \"summary\": \"...\",\n \"on_call_user\": \"...\",\n \"channel_name\": \"...\",\n \"file_ids\": [\"x4t1k...\", \"9bzqm...\"]}" }, "position": [ -48, 416 ] }, { "id": "85a3602f-4211-42cd-b0c0-7952e8ec3d1d", "name": "SetVars", "type": "set", "params": { "options": {}, "assignments": { "assignments": [ { "id": "d89e8819-2d33-4548-84f7-bd00854b70e8", "name": "MATTERMOST_URL", "type": "string", "value": "https:////" }, { "id": "16e9f447-4d98-4173-ba3e-95c047762068", "name": "reply_root_id", "type": "string", "value": "={{ $('ReadIncidentContext').item.json.is_thread ? $('ReadIncidentContext').item.json.thread_root_id : $('ReadIncidentContext').item.json.post_id}}" }, { "id": "e54f7764-afd8-487c-909a-68598092367b", "name": "JIRA_URL", "type": "string", "value": "https://.atlassian.net" }, { "id": "ddd796fc-2aa2-4bbd-aee3-e55dd9009276", "name": "message", "type": "string", "value": "={{ $('ReadIncidentContext').item.json.message }}" }, { "id": "66b+1234567890aa-81ee-585254f44f2d", "name": "post_id", "type": "string", "value": "={{ $('ReadIncidentContext').item.json.post_id }}" }, { "id": "804efcde-19f0-4854-8bdd-5b30faa15ac0", "name": "channel_id", "type": "string", "value": "={{ $('ReadIncidentContext').item.json.channel_id }}" }, { "id": "86dfdd6b-a138-49a9-a926-75869b10427b", "name": "on_call_user", "type": "string", "value": "={{ $('ReadIncidentContext').item.json.on_call_user }}" }, { "id": "ffd9f052-a36d-4aef-b97d-dda4281af9b3", "name": "channel_name", "type": "string", "value": "={{ $('ReadIncidentContext').item.json.channel_name }}" }, { "id": "fbebb9d0-4ea3-424c-b839-5098cbfcd129", "name": "is_thread", "type": "string", "value": "={{ $('ReadIncidentContext').item.json.is_thread }}" } ] }, "includeOtherFields": true }, "position": [ 944, 416 ] }, { "id": "a20b8b36-a5fe-454c-9df9-a08faf10dba8", "name": "AI Agent", "type": "llm", "params": { "text": "=Prepare a task as requested by {{ $json.user_name }} in channel {{ $json.channel_name }} {{ $json.is_thread ? ' (message is in thread, thread_root_id=' + $json.thread_root_id + ' — read the story through Mattermost get_thread first)' : '' }}\n\n{{ $json.message }}{{ $json.attachments_context && $json.attachments_context.trim().length > 0 ? '\\n\\nAdditional information from attachments:\\n' + $json.attachments_context : '' }}", "prompt": "=Prepare a task as requested by {{ $json.user_name }} in channel {{ $json.channel_name }} {{ $json.is_thread ? ' (message is in thread, thread_root_id=' + $json.thread_root_id + ' — read the story through Mattermost get_thread first)' : '' }}\n\n{{ $json.message }}{{ $json.attachments_context && $json.attachments_context.trim().length > 0 ? '\\n\\nAdditional information from attachments:\\n' + $json.attachments_context : '' }}", "options": { "maxIterations": 15, "systemMessage": "=You're a professional task manager. You're proficient in setting tasks for DevOps, SRE engineers.\nAt the user's request, you need to prepare the data to create a task in Jira. You don't need to create the task yourself.\nDetermine the `subject area`; what does the task relate to? Use the Qdrant Vector Store tool to learn more about our infrastructure when needed.\nDefine the action that needs to be performed: create something, delete something, change something.\nThis could be a database, a cluster, a logging system, or a monitoring system.\n\nDetermine the `direction` and choose exactly one:\n- HighAvailability — availability of our services\n- Performance — performance of our services\n- Security — information security\n- CostManagement — cost management\n- Operations — routine operational activities\n- Observability — monitoring, logs, traces, alerts\n\n## Message context\n\nThe request can come either from a main channel message or from a thread reply.\n\n- channel_id: {{ $json.channel_id }}\n- post_id: {{ $json.post_id }}\n- is_thread: {{ $json.is_thread }}\n- thread_root_id: {{ $json.thread_root_id }}\n\nIf `is_thread` is true, the user's question is part of an ongoing thread.\nBefore investigating, call Mattermost `get_thread` with `thread_root_id`\nexactly ONCE to load the full conversation. Earlier messages in the thread\noften contain critical context: links to failing jobs, error logs, repository\nnames, previous hypotheses, or clarifications from the user.\n\nYou can use the HTTP Request tool to probe endpoints when the request is about availability or connectivity.\n\n## Response format\nReturn a single JSON object with EXACTLY these fields — no markdown, no code fences, no extra text:\n{\"summary\":\": \",\"description\":\"<1-3 sentences>\",\"label\":\"\"}\n\n## Example 1\nRequest: create a small Postgres database for the checkout API\nOutput:\n{\"summary\":\"Postgres: create database for checkout API\",\"description\":\"We need to create small Postgres instances in the stage and prod environments for the checkout backend service.\",\"label\":\"Operations\"}\n\n## Example 2\nRequest: please enable gRPC on the realtime gateway and add the URL to Vault\nOutput:\n{\"summary\":\"Realtime gateway: enable gRPC\",\"description\":\"Need to enable gRPC on the realtime gateway and add the service URL to HashiCorp Vault.\",\"label\":\"Operations\"}\n\n## Example 3\nRequest: Need to change logs collector promtail to alloy. Create a task\nOutput:\n{\"summary\":\"Logging: replace promtail with alloy\",\"description\":\"Need to change logs collector promtail to alloy.\",\"label\":\"Observability\"}\n\n\n## Rules\n* Describe the problem in formal language, even if it is not so in the initial message" }, "provider": "openai", "promptType": "define", "approximated": true, "systemMessage": "=You're a professional task manager. You're proficient in setting tasks for DevOps, SRE engineers.\nAt the user's request, you need to prepare the data to create a task in Jira. You don't need to create the task yourself.\nDetermine the `subject area`; what does the task relate to? Use the Qdrant Vector Store tool to learn more about our infrastructure when needed.\nDefine the action that needs to be performed: create something, delete something, change something.\nThis could be a database, a cluster, a logging system, or a monitoring system.\n\nDetermine the `direction` and choose exactly one:\n- HighAvailability — availability of our services\n- Performance — performance of our services\n- Security — information security\n- CostManagement — cost management\n- Operations — routine operational activities\n- Observability — monitoring, logs, traces, alerts\n\n## Message context\n\nThe request can come either from a main channel message or from a thread reply.\n\n- channel_id: {{ $json.channel_id }}\n- post_id: {{ $json.post_id }}\n- is_thread: {{ $json.is_thread }}\n- thread_root_id: {{ $json.thread_root_id }}\n\nIf `is_thread` is true, the user's question is part of an ongoing thread.\nBefore investigating, call Mattermost `get_thread` with `thread_root_id`\nexactly ONCE to load the full conversation. Earlier messages in the thread\noften contain critical context: links to failing jobs, error logs, repository\nnames, previous hypotheses, or clarifications from the user.\n\nYou can use the HTTP Request tool to probe endpoints when the request is about availability or connectivity.\n\n## Response format\nReturn a single JSON object with EXACTLY these fields — no markdown, no code fences, no extra text:\n{\"summary\":\": \",\"description\":\"<1-3 sentences>\",\"label\":\"\"}\n\n## Example 1\nRequest: create a small Postgres database for the checkout API\nOutput:\n{\"summary\":\"Postgres: create database for checkout API\",\"description\":\"We need to create small Postgres instances in the stage and prod environments for the checkout backend service.\",\"label\":\"Operations\"}\n\n## Example 2\nRequest: please enable gRPC on the realtime gateway and add the URL to Vault\nOutput:\n{\"summary\":\"Realtime gateway: enable gRPC\",\"description\":\"Need to enable gRPC on the realtime gateway and add the service URL to HashiCorp Vault.\",\"label\":\"Operations\"}\n\n## Example 3\nRequest: Need to change logs collector promtail to alloy. Create a task\nOutput:\n{\"summary\":\"Logging: replace promtail with alloy\",\"description\":\"Need to change logs collector promtail to alloy.\",\"label\":\"Observability\"}\n\n\n## Rules\n* Describe the problem in formal language, even if it is not so in the initial message" }, "position": [ 1200, 416 ] }, { "id": "271b834e-0881-4016-9002-6529df935d6f", "name": "OpenRouter Chat Model", "type": "llm", "params": { "model": "openai/gpt-5.3-codex", "options": { "temperature": 0.2, "responseFormat": "json_object" }, "provider": "openai", "temperature": 0.2 }, "position": [ 1136, 576 ] }, { "id": "e9d25b17-61f6-4be2-968b-7a5607026855", "name": "HTTP Request", "type": "http", "params": { "url": "={{ /*n8n-auto-generated-fromAI-override*/ $fromAI('URL', ``, 'string') }}", "options": {}, "toolDescription": "Makes an HTTP request and returns the response data. For checking network issues or probing endpoints." }, "position": [ 1344, 800 ] }, { "id": "473df4d2-a9ba-4bb0-81ff-1da25d3e46e1", "name": "Qdrant Vector Store", "type": "passthrough", "params": { "mode": "noOp", "options": {}, "toolDescription": "Use it to learn more about our infrastructure, alert rules and runbooks.", "qdrantCollection": { "__rl": true, "mode": "list", "value": "", "cachedResultName": "" } }, "position": [ 1472, 800 ] }, { "id": "db631503-b383-4299-a149-28d75642f91f", "name": "Embeddings Google Gemini", "type": "passthrough", "params": { "mode": "noOp", "modelName": "models/gemini-embedding-2-preview" }, "position": [ 1472, 976 ] }, { "id": "1b7f0641-f611-4321-ba25-d93c6f46b844", "name": "Parse Agent Output", "type": "code", "params": { "jsCode": "// Parse AI Agent output (raw JSON or inside a markdown code fence),\n// merge with context from Read Incident Context and SetVars,\n// and build a Jira payload including a Mattermost permalink to the source message.\n\nconst raw = $input.first().json.output ?? '';\nconst ctx = $('ReadIncidentContext').first().json;\nconst vars = $('SetVars').first().json;\n\n// Strip surrounding markdown code fences if present\nconst cleaned = String(raw)\n .replace(/^\\s*```(?:json)?\\s*/i, '')\n .replace(/\\s*```\\s*$/i, '')\n .trim();\n\nlet parsed;\ntry {\n parsed = JSON.parse(cleaned);\n} catch (err) {\n throw new Error(`Failed to parse AI Agent output as JSON: ${err.message}. Raw output was: ${raw}`);\n}\n\n// Build a Mattermost permalink to the thread where the request was posted\nconst mattermostBase = (vars.MATTERMOST_URL || '').replace(/\\/+$/, '');\nconst threadLink = ctx.post_id ? `${mattermostBase}/${ctx.post_id}` : null;\n\n// Compose source block appended to the Jira description\nconst sourceLines = [\n '',\n '---',\n `*Source:* @${ctx.user_name || 'unknown'} in channel #${ctx.channel_name || 'unknown'}`,\n];\nif (threadLink) sourceLines.push(`*Message link:* ${threadLink}`);\nif (ctx.summary) sourceLines.push(`*Classification:* ${ctx.summary} (confidence: ${ctx.confidence ?? 'n/a'})`);\n\nconst description = `${parsed.description || ''}\\n${sourceLines.join('\\n')}`;\n\nreturn [{\n json: {\n summary: parsed.summary || '(no summary)',\n description,\n label: parsed.label || 'Operations',\n user_name: ctx.user_name,\n on_call_user: ctx.on_call_user,\n channel_id: ctx.channel_id,\n post_id: ctx.post_id,\n thread_link: threadLink,\n agent_raw_output: raw,\n },\n}];" }, "position": [ 1648, 416 ] }, { "id": "da526997-7792-4d84-a0b2-b4cb8d9b2ef9", "name": "Create an issue", "type": "integration", "params": { "project": { "__rl": true, "mode": "list", "value": "", "cachedResultName": "" }, "service": "jira", "summary": "={{ $json.summary }}", "issueType": { "__rl": true, "mode": "list", "value": "", "cachedResultName": "" }, "operation": "read", "additionalFields": { "labels": "={{ [$json.label] }}", "description": "={{ $json.description }}", "componentIds": [ "" ] } }, "position": [ 1824, 416 ] }, { "id": "62d40c47-6da6-440e-9621-1e9e853e058d", "name": "Post a message", "type": "integration", "params": { "message": "=✅ Created issue [{{ $('Create an issue').first().json.key }}]({{ $('SetVars').first().json.JIRA_URL }}/browse/{{ $('Create an issue').first().json.key }}): {{ $('Parse Agent Output').first().json.summary }}", "service": "mattermost", "channelId": "={{ $('Parse Agent Output').first().json.channel_id }}", "operation": "run", "attachments": [], "otherOptions": { "root_id": "={{ $('SetVars').first().json.reply_root_id }}" }, "__originalType": "n8n-nodes-base.mattermost" }, "position": [ 2032, 416 ] }, { "id": "ca58e161-67a4-4c9c-a60f-228d6da4698c", "name": "Mattermost", "type": "integration", "params": { "include": "selected", "options": {}, "service": "mcp", "operation": "read", "endpointUrl": "/mattermost/mcp", "includeTools": [ "get_channel_by_name", "list_public_channels", "get_file_info", "get_channel_messages", "search_messages", "get_thread", "get_me", "get_user", "get_user_by_username" ] }, "position": [ 1216, 800 ] }, { "id": "a2406fd5-fb89-46e1-a7f2-db40396945a7", "name": "Edit Fields", "type": "set", "params": { "options": {}, "assignments": { "assignments": [ { "id": "632595c9-f4cd-43f7-916c-5ed9bdf51db4", "name": "attachments_context", "type": "string", "value": "" }, { "id": "7ad957b3-1d49-47ae-b2d6-1abf5ced6b08", "name": "attachments_count", "type": "string", "value": "0" } ] } }, "position": [ 608, 560 ] }, { "id": "93828410-1aa3-4d06-9b45-e7a4537b5f3b", "name": "Merge", "type": "merge", "params": {}, "position": [ 768, 416 ] }, { "id": "780ddba7-3663-425a-9b2a-bd066aa0a5a3", "name": "ReadIncidentContext", "type": "code", "params": { "jsCode": "// Read the classification object passed from the parent classifier workflow.\n// Available fields: message, post_id, channel_id, user_name,\n// channel_name, category, confidence, summary, on_call_user.\n\nconst data = $input.first().json;\n\nconsole.log(`Task received: ${data.summary}`);\nconsole.log(`From: ${data.user_name} in #${data.channel_name}`);\nconsole.log(`Confidence: ${data.confidence}`);\n\nreturn [{ json: data }];" }, "position": [ 144, 416 ] }, { "id": "4b349210-7f35-4dc0-905f-081d945bb8a8", "name": "Call 'attachmentsAnalyzer'", "type": "passthrough", "params": { "mode": "noOp", "options": {}, "workflowId": { "__rl": true, "mode": "list", "value": "", "cachedResultUrl": "/workflow/", "cachedResultName": "" }, "workflowInputs": { "value": { "file_ids": "={{ $json.file_ids }}" }, "schema": [ { "id": "file_ids", "type": "array", "display": true, "removed": false, "required": false, "displayName": "file_ids", "defaultMatch": false, "canBeUsedToMatch": true } ], "mappingMode": "defineBelow", "matchingColumns": [ "file_ids" ], "attemptToConvertTypes": false, "convertFieldsToString": true } }, "position": [ 416, 416 ] } ] }

Convert DevOps Mattermost requests into Jira tasks with OpenRouter GPT and Gemini