{ "id": "mfj3y9MNPLaxkcwU", "meta": { "templateCredsSetupCompleted": true }, "name": "attachmentsAnalyzer", "tags": [], "nodes": [ { "id": "e342262d-24d0-4d50-bf5c-4b6df19561e2", "name": "About", "type": "n8n-nodes-base.stickyNote", "position": [ 640, -192 ], "parameters": { "color": 5, "width": 1336, "height": 544, "content": "## Overview\n\nSubworkflow for analyzing Mattermost attachments before passing them to the AI Agent.\n\n## Per-attachment behavior\n- **Image** (mime starts with `image/`, size \u2264 `MAX_IMAGE_SIZE_BYTES`): downloaded and analyzed by OpenAI vision model.\n- **Text small** (text-like by mime/extension, size \u2264 `MAX_TEXT_SIZE_BYTES`): downloaded as plain text and inlined into the context block (with line count check \u2264 `MAX_TEXT_LINES`).\n- **Text too large** (text-like, exceeds byte or line limit): replaced with a \"too big\" marker so the agent asks the user for the relevant excerpt.\n- **Other** (binaries, oversized images, unknown): replaced with a generic \"not analyzed\" marker.\n\n## Setup checklist\n1. Set `MATTERMOST_BASE_URL` in the Config node to your Mattermost origin (`https://hostname`; no trailing slash).\n2. Create an `HTTP Header Auth` credential called \"Mattermost PAT\" with:\n - Name: `Authorization`\n - Value: `Bearer `\n3. Attach this credential to all three HTTP Request nodes (Get file info, Download file, Download text file).\n4. Attach your OpenAI credential to the \"OpenAI: Analyze Image\" node.\n5. Tune in Config: `VISION_MODEL` (OpenAI vision model id; the Analyze Image node reads it from Config), `MAX_IMAGE_SIZE_BYTES`, `MAX_TEXT_SIZE_BYTES`, `MAX_TEXT_LINES`.\n\n## Notes\n- The IF guard short-circuits when `file_ids` is empty so the parent workflow can call this subworkflow unconditionally.\n- Classification by mime + extension catches `.log` files served as `application/octet-stream`.\n- Final output is always a single item with the same shape \u2014 safe to consume in the parent." }, "typeVersion": 1 }, { "id": "bede9898-e490-496a-bf78-2703c31292e5", "name": "When Executed by Another Workflow", "type": "n8n-nodes-base.executeWorkflowTrigger", "position": [ 704, 656 ], "parameters": { "inputSource": "jsonExample", "jsonExample": "{\n \"file_ids\": [\"x4t1k7example1\", \"9bzqmexample2\"]\n}" }, "typeVersion": 1.1 }, { "id": "2cbd0079-0953-4cfd-ba09-89b130e3e065", "name": "Config", "type": "n8n-nodes-base.set", "position": [ 896, 656 ], "parameters": { "options": {}, "assignments": { "assignments": [ { "id": "c1f+1234567890-+1234567890", "name": "MATTERMOST_BASE_URL", "type": "string", "value": "https://" }, { "id": "c1f+1234567890-+1234567890", "name": "VISION_MODEL", "type": "string", "value": "gpt-4o-mini" }, { "id": "c1f+1234567890-+1234567890", "name": "MAX_IMAGE_SIZE_BYTES", "type": "number", "value": 20971520 }, { "id": "c1f+1234567890-+1234567890", "name": "MAX_TEXT_SIZE_BYTES", "type": "number", "value": 65536 }, { "id": "c1f+1234567890-+1234567890", "name": "MAX_TEXT_LINES", "type": "number", "value": 300 } ] }, "includeOtherFields": true }, "typeVersion": 3.4 }, { "id": "50298b55-4fbf-4052-8859-3ae9f1d73c23", "name": "Has attachments?", "type": "n8n-nodes-base.if", "position": [ 1088, 656 ], "parameters": { "options": {}, "conditions": { "options": { "version": 2, "leftValue": "", "caseSensitive": true, "typeValidation": "strict" }, "combinator": "and", "conditions": [ { "id": "if+1234567890-+1234567890", "operator": { "type": "number", "operation": "gt" }, "leftValue": "={{ ($json.file_ids || []).length }}", "rightValue": 0 } ] } }, "typeVersion": 2.2 }, { "id": "7369a49e-7f79-4082-a07b-1ef09b87f6f4", "name": "Empty result", "type": "n8n-nodes-base.set", "position": [ 1296, 912 ], "parameters": { "options": {}, "assignments": { "assignments": [ { "id": "se+1234567890-+1234567890", "name": "attachments_context", "type": "string", "value": "" }, { "id": "se+1234567890-+1234567890", "name": "attachments_count", "type": "number", "value": 0 } ] } }, "typeVersion": 3.4 }, { "id": "08c9ba1a-6413-4a5c-a129-6fa456d5e907", "name": "Split file_ids", "type": "n8n-nodes-base.code", "position": [ 1328, 464 ], "parameters": { "jsCode": "// Split incoming file_ids array into one item per file_id.\n// The IF guard upstream guarantees at least one id; defensive check kept.\n\nconst data = $input.first().json;\nconst ids = Array.isArray(data.file_ids) ? data.file_ids : [];\n\nreturn ids.map((file_id, idx) => ({\n json: {\n file_id,\n file_index: idx\n }\n}));" }, "typeVersion": 2 }, { "id": "1e0aae90-4e03-4f2d-9639-0c15bbd7aff8", "name": "Get file info", "type": "n8n-nodes-base.httpRequest", "position": [ 1552, 464 ], "parameters": { "url": "={{ $('Config').first().json.MATTERMOST_BASE_URL }}/api/v4/files/{{ $json.file_id }}/info", "options": {}, "authentication": "genericCredentialType", "genericAuthType": "httpHeaderAuth" }, "credentials": { "httpHeaderAuth": { "name": "" } }, "typeVersion": 4.2 }, { "id": "1b35f66f-fef6-4e36-9afe-4bd2408cda3c", "name": "Classify file", "type": "n8n-nodes-base.code", "position": [ 1760, 464 ], "parameters": { "mode": "runOnceForEachItem", "jsCode": "// Classify the attachment by mime type, file extension, and size.\n// Adds a `category` field used by the downstream Switch node:\n// - 'image' image, fits MAX_IMAGE_SIZE_BYTES \u2192 vision pipeline\n// - 'text_small' text-like, fits MAX_TEXT_SIZE_BYTES \u2192 inline content\n// - 'text_too_large' text-like but exceeds MAX_TEXT_SIZE_BYTES \u2192 \"too big\" marker\n// - 'unsupported' binaries, oversized images, unknown formats\n//\n// Detecting text by extension is intentional: Mattermost often serves `.log`\n// (and other dev-related files) as `application/octet-stream`.\n\nconst fileInfo = $input.item.json;\nconst config = $('Config').first().json;\n\nconst mimeType = (fileInfo.mime_type || '').toLowerCase();\nconst fileName = fileInfo.name || '';\nconst size = Number(fileInfo.size) || 0;\n\nconst dotIdx = fileName.lastIndexOf('.');\nconst ext = dotIdx >= 0 ? fileName.slice(dotIdx + 1).toLowerCase() : '';\n\nconst TEXT_EXTENSIONS = new Set([\n // Logs & plain text\n 'log', 'txt', 'out', 'err',\n // Structured data\n 'json', 'ndjson', 'jsonl', 'yaml', 'yml', 'xml', 'csv', 'tsv', 'toml',\n // Docs & configs\n 'md', 'markdown', 'rst', 'conf', 'cfg', 'ini', 'env', 'properties',\n // Shell & scripts\n 'sh', 'bash', 'zsh', 'fish', 'ps1',\n // Programming languages\n 'py', 'js', 'mjs', 'cjs', 'ts', 'jsx', 'tsx',\n 'go', 'rs', 'java', 'kt', 'rb', 'php', 'pl', 'lua',\n 'c', 'h', 'cpp', 'hpp', 'cc', 'cs', 'm', 'mm',\n // Infra & DevOps\n 'tf', 'hcl', 'tfvars', 'sql', 'graphql', 'gql', 'dockerfile',\n // Web\n 'html', 'htm', 'css', 'scss', 'sass', 'less', 'vue', 'svelte',\n // Patches\n 'patch', 'diff'\n]);\n\nconst isImage = mimeType.startsWith('image/');\nconst isTextMime =\n mimeType.startsWith('text/') ||\n mimeType === 'application/json' ||\n mimeType === 'application/ld+json' ||\n mimeType === 'application/xml' ||\n mimeType === 'application/x-yaml' ||\n mimeType === 'application/yaml' ||\n mimeType === 'application/x-sh' ||\n mimeType === 'application/javascript' ||\n mimeType === 'application/x-ndjson' ||\n mimeType === 'application/toml';\nconst isTextLike = isTextMime || TEXT_EXTENSIONS.has(ext);\n\nlet category;\nif (isImage && size <= config.MAX_IMAGE_SIZE_BYTES) {\n category = 'image';\n} else if (isTextLike && size <= config.MAX_TEXT_SIZE_BYTES) {\n category = 'text_small';\n} else if (isTextLike) {\n category = 'text_too_large';\n} else {\n category = 'unsupported';\n}\n\nreturn {\n json: {\n ...fileInfo,\n file_extension: ext,\n is_text_like: isTextLike,\n is_image: isImage,\n category\n }\n};" }, "typeVersion": 2 }, { "id": "d7abba2a-4351-4bae-8ea5-cc1c45335d9c", "name": "Route by category", "type": "n8n-nodes-base.switch", "position": [ 1968, 432 ], "parameters": { "rules": { "values": [ { "outputKey": "image", "conditions": { "options": { "version": 2, "leftValue": "", "caseSensitive": true, "typeValidation": "strict" }, "combinator": "and", "conditions": [ { "id": "sw+1234567890-+1234567890", "operator": { "type": "string", "operation": "equals" }, "leftValue": "={{ $json.category }}", "rightValue": "image" } ] }, "renameOutput": true }, { "outputKey": "text_small", "conditions": { "options": { "version": 2, "leftValue": "", "caseSensitive": true, "typeValidation": "strict" }, "combinator": "and", "conditions": [ { "id": "sw+1234567890-+1234567890", "operator": { "type": "string", "operation": "equals" }, "leftValue": "={{ $json.category }}", "rightValue": "text_small" } ] }, "renameOutput": true }, { "outputKey": "text_too_large", "conditions": { "options": { "version": 2, "leftValue": "", "caseSensitive": true, "typeValidation": "strict" }, "combinator": "and", "conditions": [ { "id": "sw+1234567890-+1234567890", "operator": { "type": "string", "operation": "equals" }, "leftValue": "={{ $json.category }}", "rightValue": "text_too_large" } ] }, "renameOutput": true } ] }, "options": { "fallbackOutput": "extra", "renameFallbackOutput": "unsupported" } }, "typeVersion": 3.2 }, { "id": "51e2994b-61c6-4797-b521-616f6005f53d", "name": "Download file", "type": "n8n-nodes-base.httpRequest", "position": [ 2192, 160 ], "parameters": { "url": "={{ $('Config').first().json.MATTERMOST_BASE_URL }}/api/v4/files/{{ $json.id }}", "options": { "response": { "response": { "responseFormat": "file" } } }, "authentication": "genericCredentialType", "genericAuthType": "httpHeaderAuth" }, "credentials": { "httpHeaderAuth": { "name": "" } }, "typeVersion": 4.2 }, { "id": "720bd88c-5091-4b38-a58f-488106b3b067", "name": "OpenAI: Analyze Image", "type": "@n8n/n8n-nodes-langchain.openAi", "position": [ 2416, 160 ], "parameters": { "text": "You are a DevOps screenshot analyst. Produce a concise structured description in English of the attached image.\n\nOutput sections (skip a section if not applicable):\n1. source_system \u2014 which tool/system the screenshot is from. Choose from: GitHub Actions, GitLab CI, Jenkins, TeamCity, Grafana, Sentry, Loki, Prometheus, Mattermost, Slack, Jira, Kubernetes Dashboard, terminal/shell, IDE (VSCode/JetBrains), browser DevTools, generic web page, unknown.\n2. visible_text \u2014 verbatim transcript of error messages, stack traces, command output, status codes, job names. Preserve exact wording and casing. Use code fences for multi-line blocks.\n3. key_signals \u2014 bullet list of what matters for an SRE: failed step name, exit code, service/pod/namespace, timestamp, severity, status indicator color.\n4. summary \u2014 1-2 sentences in plain English on what the screenshot shows.\n\nBe terse. No filler, no greetings, no markdown headings beyond what is asked. If the image is not technical (meme, photo of a person, decorative), say so in summary and skip other sections.", "modelId": { "__rl": true, "mode": "id", "value": "={{ $('Config').first().json.VISION_MODEL }}" }, "options": { "detail": "auto" }, "resource": "image", "inputType": "=base64", "operation": "analyze" }, "credentials": { "openAiApi": { "name": "" } }, "typeVersion": 1.8 }, { "id": "563c042b-8359-4eed-b5fa-32ea62da3a19", "name": "Format image description", "type": "n8n-nodes-base.code", "position": [ 2640, 160 ], "parameters": { "mode": "runOnceForEachItem", "jsCode": "// runOnceForEachItem \u2014 pairedItem propagation lets us reach back to\n// \"Classify file\" by item index and recover the original file metadata.\n\nconst llmOut = $input.item.json;\nconst fileInfo = $('Classify file').item.json;\n\n// OpenAI Analyze Image returns the model text under different fields\n// depending on n8n version. Try common shapes in order.\nlet description =\n (typeof llmOut.content === 'string' && llmOut.content) ||\n (typeof llmOut.message?.content === 'string' && llmOut.message.content) ||\n (Array.isArray(llmOut.content) && llmOut.content.map(c => c.text || '').join('\\n')) ||\n llmOut.text ||\n '[vision model returned no content]';\n\nif (typeof description !== 'string') {\n description = JSON.stringify(description);\n}\n\nreturn {\n json: {\n file_id: fileInfo.id,\n file_name: fileInfo.name,\n mime_type: fileInfo.mime_type,\n size: fileInfo.size,\n is_image: true,\n is_text: false,\n description\n }\n};" }, "typeVersion": 2 }, { "id": "9b8f65b0-012a-4d17-a53c-65931b5e7cf2", "name": "Download text file", "type": "n8n-nodes-base.httpRequest", "position": [ 2192, 352 ], "parameters": { "url": "={{ $('Config').first().json.MATTERMOST_BASE_URL }}/api/v4/files/{{ $json.id }}", "options": { "response": { "response": { "responseFormat": "text" } } }, "authentication": "genericCredentialType", "genericAuthType": "httpHeaderAuth" }, "credentials": { "httpHeaderAuth": { "name": "" } }, "typeVersion": 4.2 }, { "id": "eb150e5f-ddb7-4387-b8c8-55ffd741f22d", "name": "Format text content", "type": "n8n-nodes-base.code", "position": [ 2416, 352 ], "parameters": { "mode": "runOnceForEachItem", "jsCode": "// runOnceForEachItem. HTTP Request (responseFormat=text) returned the body\n// as a string in $json.data. File metadata is recovered via pairedItem from\n// \"Classify file\".\n//\n// Two-tier size check:\n// tier 1 (bytes): already enforced by \"Classify file\" + Switch routing.\n// tier 2 (lines): re-checked here for files that fit byte budget but\n// contain unusually long lines (e.g. minified JSON).\n\nconst fileInfo = $('Classify file').item.json;\nconst config = $('Config').first().json;\nconst text = typeof $json.data === 'string' ? $json.data : '';\n\nconst lineCount = text.length === 0 ? 0 : text.split(/\\r\\n|\\r|\\n/).length;\nconst maxLines = Number(config.MAX_TEXT_LINES) || 300;\n\nif (lineCount > maxLines) {\n return {\n json: {\n file_id: fileInfo.id,\n file_name: fileInfo.name,\n mime_type: fileInfo.mime_type,\n size: fileInfo.size,\n is_image: false,\n is_text: true,\n description: `Attachment ${fileInfo.name} is too large to inline: ${fileInfo.size} bytes / ${lineCount} lines (limit: ${maxLines} lines). Ask the user to share the relevant excerpt.`\n }\n };\n}\n\nreturn {\n json: {\n file_id: fileInfo.id,\n file_name: fileInfo.name,\n mime_type: fileInfo.mime_type,\n size: fileInfo.size,\n is_image: false,\n is_text: true,\n description: `Text content of ${fileInfo.name} (${fileInfo.size} bytes, ${lineCount} lines):\\n\\n\\`\\`\\`\\n${text}\\n\\`\\`\\``\n }\n};" }, "typeVersion": 2 }, { "id": "bf4dfecb-8585-4252-871f-dfee7970dc04", "name": "Text too large marker", "type": "n8n-nodes-base.set", "position": [ 2192, 576 ], "parameters": { "options": {}, "assignments": { "assignments": [ { "id": "tl+1234567890-+1234567890", "name": "file_id", "type": "string", "value": "={{ $json.id }}" }, { "id": "tl+1234567890-+1234567890", "name": "file_name", "type": "string", "value": "={{ $json.name }}" }, { "id": "tl+1234567890-+1234567890", "name": "mime_type", "type": "string", "value": "={{ $json.mime_type }}" }, { "id": "tl+1234567890-+1234567890", "name": "size", "type": "number", "value": "={{ $json.size }}" }, { "id": "tl+1234567890-+1234567890", "name": "is_image", "type": "boolean", "value": false }, { "id": "tl+1234567890-+1234567890", "name": "is_text", "type": "boolean", "value": true }, { "id": "tl+1234567890-+1234567890", "name": "description", "type": "string", "value": "=Attachment {{ $json.name }} is a text file but too large to inline: {{ $json.size }} bytes (limit: {{ $('Config').first().json.MAX_TEXT_SIZE_BYTES }} bytes). Ask the user to share the relevant excerpt." } ] } }, "typeVersion": 3.4 }, { "id": "63ec86c3-eb4b-4da3-bbc7-c80293618a44", "name": "Non-image marker", "type": "n8n-nodes-base.set", "position": [ 2192, 784 ], "parameters": { "options": {}, "assignments": { "assignments": [ { "id": "ni+1234567890-+1234567890", "name": "file_id", "type": "string", "value": "={{ $json.id }}" }, { "id": "ni+1234567890-+1234567890", "name": "file_name", "type": "string", "value": "={{ $json.name }}" }, { "id": "ni+1234567890-+1234567890", "name": "mime_type", "type": "string", "value": "={{ $json.mime_type }}" }, { "id": "ni+1234567890-+1234567890", "name": "size", "type": "number", "value": "={{ $json.size }}" }, { "id": "ni+1234567890-+1234567890", "name": "is_image", "type": "boolean", "value": "={{ ($json.mime_type || '').startsWith('image/') }}" }, { "id": "ni+1234567890-+1234567890", "name": "is_text", "type": "boolean", "value": false }, { "id": "ni+1234567890-+1234567890", "name": "description", "type": "string", "value": "=Attachment {{ $json.name }} not analyzed (mime: {{ $json.mime_type }}, size: {{ $json.size }} bytes). Either unsupported binary format or exceeds size limit." } ] } }, "typeVersion": 3.4 }, { "id": "f716620e-b4d8-45fb-b080-aa311dc79d43", "name": "Merge", "type": "n8n-nodes-base.merge", "position": [ 2864, 528 ], "parameters": { "numberInputs": 4 }, "typeVersion": 3.2 }, { "id": "5cbc7aa3-6eae-4e39-8363-e7fa7ca8efb9", "name": "Aggregate", "type": "n8n-nodes-base.aggregate", "position": [ 3072, 560 ], "parameters": { "options": {}, "aggregate": "aggregateAllItemData", "destinationFieldName": "attachments" }, "typeVersion": 1 }, { "id": "3de99b38-8fec-432f-9f9c-2fcfcff19d5b", "name": "Build attachments_context", "type": "n8n-nodes-base.code", "position": [ 3264, 560 ], "parameters": { "jsCode": "// Aggregate produced a single item with `attachments: [...]`.\n// Build a human-readable block for the AI Agent system prompt.\n// Output shape mirrors the \"Empty result\" Set node so the parent workflow\n// can consume the output uniformly.\n\nconst attachments = $input.first().json.attachments || [];\nconst valid = attachments.filter(a => a && a.file_id);\n\nif (valid.length === 0) {\n return [{ json: { attachments_context: '', attachments_count: 0 } }];\n}\n\nconst blocks = valid.map((a, i) => {\n let tag;\n if (a.is_image) tag = 'image';\n else if (a.is_text) tag = 'text';\n else tag = a.mime_type || 'file';\n\n const header = `Attachment ${i + 1}: ${a.file_name} [${tag}]`;\n return `${header}\\n${a.description}`;\n});\n\nreturn [{\n json: {\n attachments_context: blocks.join('\\n\\n---\\n\\n'),\n attachments_count: valid.length\n }\n}];" }, "typeVersion": 2 }, { "id": "70b6af74-e8da-464a-b47c-ac1426a65865", "name": "Sticky Note", "type": "n8n-nodes-base.stickyNote", "position": [ 640, 576 ], "parameters": { "width": 576, "height": 480, "content": "# Input chain\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n```json\n{ \"file_ids\": [\"id1\", \"id2\", ...] }\n```\n" }, "typeVersion": 1 }, { "id": "c830d6bb-75fb-4004-8a15-2b81efc5f44b", "name": "Sticky Note1", "type": "n8n-nodes-base.stickyNote", "position": [ 3008, 480 ], "parameters": { "width": 528, "height": 512, "content": "## Output chain\n\n\n\n\n\n\n\n\n\n\n\n\n```json\n{\n \"attachments_context\": \"...human-readable block...\",\n \"attachments_count\": 2\n}\n```" }, "typeVersion": 1 }, { "id": "91fa12c8-0d2f-4880-88c8-b011a67a40fe", "name": "Sticky Note2", "type": "n8n-nodes-base.stickyNote", "position": [ 2160, 48 ], "parameters": { "color": 6, "width": 624, "height": 944, "content": "## Attachment analyzer\nAccepts images and text files" }, "typeVersion": 1 } ], "active": true, "settings": { "binaryMode": "separate", "executionOrder": "v1" }, "versionId": "8ce4e88e-800c-4a8d-b9a1-bdb69f3d86bf", "connections": { "Merge": { "main": [ [ { "node": "Aggregate", "type": "main", "index": 0 } ] ] }, "Config": { "main": [ [ { "node": "Has attachments?", "type": "main", "index": 0 } ] ] }, "Aggregate": { "main": [ [ { "node": "Build attachments_context", "type": "main", "index": 0 } ] ] }, "Classify file": { "main": [ [ { "node": "Route by category", "type": "main", "index": 0 } ] ] }, "Download file": { "main": [ [ { "node": "OpenAI: Analyze Image", "type": "main", "index": 0 } ] ] }, "Get file info": { "main": [ [ { "node": "Classify file", "type": "main", "index": 0 } ] ] }, "Split file_ids": { "main": [ [ { "node": "Get file info", "type": "main", "index": 0 } ] ] }, "Has attachments?": { "main": [ [ { "node": "Split file_ids", "type": "main", "index": 0 } ], [ { "node": "Empty result", "type": "main", "index": 0 } ] ] }, "Non-image marker": { "main": [ [ { "node": "Merge", "type": "main", "index": 3 } ] ] }, "Route by category": { "main": [ [ { "node": "Download file", "type": "main", "index": 0 } ], [ { "node": "Download text file", "type": "main", "index": 0 } ], [ { "node": "Text too large marker", "type": "main", "index": 0 } ], [ { "node": "Non-image marker", "type": "main", "index": 0 } ] ] }, "Download text file": { "main": [ [ { "node": "Format text content", "type": "main", "index": 0 } ] ] }, "Format text content": { "main": [ [ { "node": "Merge", "type": "main", "index": 1 } ] ] }, "OpenAI: Analyze Image": { "main": [ [ { "node": "Format image description", "type": "main", "index": 0 } ] ] }, "Text too large marker": { "main": [ [ { "node": "Merge", "type": "main", "index": 2 } ] ] }, "Format image description": { "main": [ [ { "node": "Merge", "type": "main", "index": 0 } ] ] }, "When Executed by Another Workflow": { "main": [ [ { "node": "Config", "type": "main", "index": 0 } ] ] } } }