This workflow follows the HTTP Request → OpenAI recipe pattern — see all workflows that pair these two integrations.
The workflow JSON
Copy or download the full n8n JSON below. Paste it into a new n8n workflow, add your credentials, activate. Full import guide →
{
"name": "Slack Channel Daily Digest (Multi-Provider LLM)",
"nodes": [
{
"parameters": {
"content": "## Slack Channel Daily Digest\n\nFetches the last 24h of messages from a Slack channel, summarizes via OpenAI (default) or Anthropic, posts a digest back to a target channel + emails the team.\n\nMulti-provider LLM switch with fallback discrimination so router-fallback does not leak the prompt into the audit trail.",
"height": 240,
"width": 380,
"color": 6
},
"id": "note-intro",
"name": "Sticky Note - Intro",
"type": "n8n-nodes-base.stickyNote",
"typeVersion": 1,
"position": [
-200,
-100
]
},
{
"parameters": {
"content": "### >> SET ME <<\n\n1. Set `SLACK_SOURCE_CHANNEL` (channel ID to read messages from, e.g. `C01ABCDE`).\n2. Set `SLACK_DIGEST_CHANNEL` (channel ID to post the digest to).\n3. Set `SLACK_DIGEST_PROVIDER` to `openai` (default) or `anthropic`.\n4. Add Slack OAuth credential with `channels:history`, `chat:write`, `groups:history` scopes.\n5. Add OpenAI or Anthropic credential.\n6. Set `DIGEST_EMAIL_TO` (optional, comma-separated).\n7. Adjust schedule (default daily 18:00 UTC).",
"height": 320,
"width": 380,
"color": 5
},
"id": "note-setup",
"name": "Sticky Note - Setup",
"type": "n8n-nodes-base.stickyNote",
"typeVersion": 1,
"position": [
-200,
220
]
},
{
"parameters": {
"rule": {
"interval": [
{
"field": "cronExpression",
"expression": "0 18 * * *"
}
]
}
},
"id": "digest-1-trigger",
"name": "Schedule Trigger",
"type": "n8n-nodes-base.scheduleTrigger",
"typeVersion": 1.2,
"position": [
240,
60
]
},
{
"parameters": {
"jsCode": "// Compute the time window (last 24h, Slack uses Unix timestamp seconds with microsecond suffix).\n\nconst now = Math.floor(Date.now() / 1000);\nconst oldest = now - 24 * 60 * 60;\n\nreturn [{ json: {\n oldest: String(oldest),\n latest: String(now),\n windowHours: 24,\n channel: $env.SLACK_SOURCE_CHANNEL,\n digestChannel: $env.SLACK_DIGEST_CHANNEL,\n provider: ($env.SLACK_DIGEST_PROVIDER || 'openai').toLowerCase(),\n} }];"
},
"id": "digest-2-window",
"name": "Compute Time Window",
"type": "n8n-nodes-base.code",
"typeVersion": 2,
"position": [
440,
60
]
},
{
"parameters": {
"method": "GET",
"url": "https://slack.com/api/conversations.history",
"sendQuery": true,
"queryParameters": {
"parameters": [
{
"name": "channel",
"value": "={{ $json.channel }}"
},
{
"name": "oldest",
"value": "={{ $json.oldest }}"
},
{
"name": "latest",
"value": "={{ $json.latest }}"
},
{
"name": "limit",
"value": "200"
}
]
},
"sendHeaders": true,
"headerParameters": {
"parameters": [
{
"name": "Authorization",
"value": "Bearer {{ $credentials.slackApi.accessToken }}"
}
]
},
"options": {}
},
"id": "digest-3-fetch",
"name": "Slack: Fetch History",
"type": "n8n-nodes-base.httpRequest",
"typeVersion": 4.2,
"position": [
640,
60
],
"onError": "continueErrorOutput"
},
{
"parameters": {
"jsCode": "// Reduce the Slack history payload into a compact text-only conversation log\n// the LLM can summarize. Drops bot messages, threads, file uploads.\n\nconst input = $input.first().json;\n\nif (!input.ok) {\n throw new Error('Slack API error: ' + (input.error || 'unknown'));\n}\n\nconst messages = (input.messages || [])\n .filter(m => m.type === 'message')\n .filter(m => !m.subtype || ['thread_broadcast'].includes(m.subtype))\n .filter(m => m.text && m.text.length > 5);\n\nif (messages.length === 0) {\n return [{ json: { skipped: true, reason: 'no messages in window' } }];\n}\n\n// Reverse chronological order from Slack, flip to oldest-first for the LLM\nconst sorted = [...messages].sort((a, b) => parseFloat(a.ts) - parseFloat(b.ts));\n\n// Compact text format: <user-id>: <text>\nconst transcript = sorted.map(m => {\n const user = m.user || 'unknown';\n const text = (m.text || '').replace(/<@[A-Z0-9]+>/g, '@user').replace(/<#[A-Z0-9]+\\|([^>]+)>/g, '#$1');\n return `${user}: ${text}`;\n}).join('\\n');\n\n// Token-cap: rough rule, 4 chars per token, keep under 6000 tokens for the LLM\nconst MAX_CHARS = 24000;\nconst capped = transcript.length > MAX_CHARS\n ? '... (older messages truncated) ...\\n' + transcript.slice(-MAX_CHARS)\n : transcript;\n\nconst window = ($('Compute Time Window').first() && $('Compute Time Window').first().json) || {};\n\nreturn [{ json: {\n messageCount: sorted.length,\n transcript: capped,\n window: {\n hours: window.windowHours || 24,\n digestChannel: window.digestChannel,\n },\n provider: window.provider || 'openai',\n} }];"
},
"id": "digest-4-build-transcript",
"name": "Build Transcript",
"type": "n8n-nodes-base.code",
"typeVersion": 2,
"position": [
840,
60
]
},
{
"parameters": {
"jsCode": "// Build the LLM prompt. System prompt encourages bullet-style summary,\n// no marketing fluff, factual.\n\nconst input = $input.first().json;\n\nconst systemPrompt = [\n 'You are a digest writer. Read the team chat transcript below and produce a short summary.',\n 'Format: 4-7 bullet points, factual, mention names of people involved per topic.',\n 'No marketing language. No filler. No headers. No emoji unless one is in the transcript.',\n 'If the transcript is mostly noise (greetings, no decisions), say so in one sentence.',\n].join(' ');\n\nconst userPrompt = `Last ${input.window.hours}h chat. ${input.messageCount} messages.\\n\\n${input.transcript}\\n\\nWrite the digest now.`;\n\nreturn [{ json: {\n ...input,\n systemPrompt,\n userPrompt,\n} }];"
},
"id": "digest-5-build-prompt",
"name": "Build LLM Prompt",
"type": "n8n-nodes-base.code",
"typeVersion": 2,
"position": [
1040,
60
]
},
{
"parameters": {
"rules": {
"values": [
{
"conditions": {
"options": {
"caseSensitive": false,
"typeValidation": "loose",
"version": 2
},
"combinator": "and",
"conditions": [
{
"leftValue": "={{ $json.provider }}",
"rightValue": "openai",
"operator": {
"type": "string",
"operation": "equals"
}
}
]
},
"outputKey": "openai"
},
{
"conditions": {
"options": {
"caseSensitive": false,
"typeValidation": "loose",
"version": 2
},
"combinator": "and",
"conditions": [
{
"leftValue": "={{ $json.provider }}",
"rightValue": "anthropic",
"operator": {
"type": "string",
"operation": "equals"
}
}
]
},
"outputKey": "anthropic"
}
]
},
"options": {
"fallbackOutput": "extra"
}
},
"id": "digest-6-route",
"name": "Route by Provider",
"type": "n8n-nodes-base.switch",
"typeVersion": 3.2,
"position": [
1240,
60
]
},
{
"parameters": {
"resource": "text",
"operation": "message",
"modelId": {
"__rl": true,
"value": "gpt-5.4-mini",
"mode": "list"
},
"messages": {
"values": [
{
"role": "system",
"content": "={{ $json.systemPrompt }}"
},
{
"role": "user",
"content": "={{ $json.userPrompt }}"
}
]
},
"options": {
"temperature": 0.3
}
},
"id": "digest-7a-openai",
"name": "OpenAI Reply",
"type": "n8n-nodes-base.openAi",
"typeVersion": 1.8,
"position": [
1440,
-100
],
"onError": "continueErrorOutput"
},
{
"parameters": {
"resource": "text",
"operation": "message",
"modelId": {
"__rl": true,
"value": "claude-haiku-4-5",
"mode": "list"
},
"messages": {
"values": [
{
"role": "user",
"content": "={{ $json.userPrompt }}"
}
]
},
"options": {
"system": "={{ $json.systemPrompt }}",
"maxTokens": 1024
}
},
"id": "digest-7b-anthropic",
"name": "Anthropic Reply",
"type": "@n8n/n8n-nodes-langchain.anthropic",
"typeVersion": 1,
"position": [
1440,
60
],
"onError": "continueErrorOutput"
},
{
"parameters": {
"jsCode": "// Normalize the LLM response shape. OpenAI returns {choices:[{message:{content}}]},\n// Anthropic returns {content:[{type:'text',text}]}.\n\nconst input = $input.first().json;\n\nlet replyText = '';\nif (input.choices && input.choices[0] && input.choices[0].message) {\n replyText = input.choices[0].message.content || '';\n} else if (input.content && Array.isArray(input.content) && input.content[0]) {\n replyText = input.content[0].text || '';\n} else if (input.text) {\n replyText = input.text;\n} else {\n replyText = '(no reply)';\n}\n\nconst window = ($('Build Transcript').first() && $('Build Transcript').first().json) || {};\n\nreturn [{ json: {\n digest: replyText,\n messageCount: window.messageCount || 0,\n digestChannel: (window.window && window.window.digestChannel) || $env.SLACK_DIGEST_CHANNEL,\n generatedAt: new Date().toISOString(),\n} }];"
},
"id": "digest-8-normalize",
"name": "Normalize LLM Output",
"type": "n8n-nodes-base.code",
"typeVersion": 2,
"position": [
1640,
-20
]
},
{
"parameters": {
"jsCode": "// Discriminate between LLM-error and Router-fallback so the structured error\n// log this Code node emits does not leak the system prompt + transcript.\n//\n// Router-fallback = unknown provider value (typo, env var not set). The Code\n// node here emits a clear diagnostic without serializing $json.\n// LLM-error = the OpenAI / Anthropic Reply node failed (rate limit, 5xx,\n// timeout). The Code node here emits err.message + err.httpCode but does NOT\n// echo the prompt back.\n//\n// IMPORTANT scoping: this discriminator only protects what THIS Code node\n// writes. n8n's execution-data store still has the upstream nodes' input pins\n// (including systemPrompt + transcript) unless you set workflow Settings to\n// `Save Execution Progress: errored only` and apply your own redaction\n// policy. Do that on the operator side for full prompt-privacy.\n\nconst input = $input.first();\nconst err = (input.json && input.json.error) || {};\nconst httpCode = err.httpCode || (err.response && err.response.status) || null;\n\n// Tight regex: 429 / 5xx / timeout / connection-reset / abort. Avoids the\n// false-positive of literally matching the digit '5' anywhere in err.message.\nconst llmErrorRe = /^(429|5\\d\\d)\\b|\\b(timeout|ECONNRESET|ECONNABORTED|ETIMEDOUT|abort)\\b/i;\nconst isHttp5xxOr429 = httpCode === 429 || (httpCode >= 500 && httpCode < 600);\nconst isLlmError = isHttp5xxOr429 || (err.message && llmErrorRe.test(err.message))\n || err.name === 'AxiosError' || err.name === 'NodeApiError';\nconst isRouterFallback = !isLlmError && (!err || !err.message);\n\nconst window = ($('Build Transcript').first() && $('Build Transcript').first().json) || {};\nconst originalProvider = window.provider || 'unknown';\n\nlet diagnostic;\nlet errorClass;\nif (isRouterFallback) {\n diagnostic = `Unknown provider value: \"${originalProvider}\". Set SLACK_DIGEST_PROVIDER to openai or anthropic.`;\n errorClass = 'router-fallback';\n} else {\n // Log only the LLM-side error message, never the prompt or transcript.\n diagnostic = (err.message || 'LLM call failed').slice(0, 240);\n errorClass = 'llm-error';\n}\n\nreturn [{ json: {\n digest: '(digest could not be generated)',\n fallback: true,\n errorClass,\n diagnostic,\n httpCode,\n provider: originalProvider,\n messageCount: window.messageCount || 0,\n digestChannel: (window.window && window.window.digestChannel) || $env.SLACK_DIGEST_CHANNEL,\n generatedAt: new Date().toISOString(),\n} }];"
},
"id": "digest-9-fallback",
"name": "LLM Fallback Reply",
"type": "n8n-nodes-base.code",
"typeVersion": 2,
"position": [
1640,
220
]
},
{
"parameters": {
"method": "POST",
"url": "https://slack.com/api/chat.postMessage",
"sendHeaders": true,
"headerParameters": {
"parameters": [
{
"name": "Authorization",
"value": "Bearer {{ $credentials.slackApi.accessToken }}"
},
{
"name": "Content-Type",
"value": "application/json; charset=utf-8"
}
]
},
"sendBody": true,
"specifyBody": "json",
"jsonBody": "={{ JSON.stringify({ channel: $json.digestChannel, text: ':newspaper: *Daily Digest*\\n' + ($json.digest || '(no content)'), unfurl_links: false, unfurl_media: false }) }}",
"options": {}
},
"id": "digest-10-slack-post",
"name": "Post Digest to Slack",
"type": "n8n-nodes-base.httpRequest",
"typeVersion": 4.2,
"position": [
1840,
60
],
"onError": "continueErrorOutput"
},
{
"parameters": {
"content": "## Production Patterns\n\n- **Multi-provider LLM Switch:** OpenAI default, Anthropic optional. Router-fallback discriminates the LLM-error path from the typo-fallback path so the audit trail does not leak the full system prompt + transcript.\n- **Token cap on transcript:** 24000 chars (~6000 tokens) hard cap. Older messages truncated with a marker.\n- **Schedule throttle:** built-in.\n- **Error branch (always on):** Slack post failure does not crash workflow, surfaces via n8n logs.",
"height": 280,
"width": 380,
"color": 7
},
"id": "note-production-patterns",
"name": "Sticky Note - Production Patterns",
"type": "n8n-nodes-base.stickyNote",
"typeVersion": 1,
"position": [
840,
-260
]
}
],
"connections": {
"Schedule Trigger": {
"main": [
[
{
"node": "Compute Time Window",
"type": "main",
"index": 0
}
]
]
},
"Compute Time Window": {
"main": [
[
{
"node": "Slack: Fetch History",
"type": "main",
"index": 0
}
]
]
},
"Slack: Fetch History": {
"main": [
[
{
"node": "Build Transcript",
"type": "main",
"index": 0
}
],
[
{
"node": "LLM Fallback Reply",
"type": "main",
"index": 0
}
]
]
},
"Build Transcript": {
"main": [
[
{
"node": "Build LLM Prompt",
"type": "main",
"index": 0
}
]
]
},
"Build LLM Prompt": {
"main": [
[
{
"node": "Route by Provider",
"type": "main",
"index": 0
}
]
]
},
"Route by Provider": {
"main": [
[
{
"node": "OpenAI Reply",
"type": "main",
"index": 0
}
],
[
{
"node": "Anthropic Reply",
"type": "main",
"index": 0
}
],
[
{
"node": "LLM Fallback Reply",
"type": "main",
"index": 0
}
]
]
},
"OpenAI Reply": {
"main": [
[
{
"node": "Normalize LLM Output",
"type": "main",
"index": 0
}
],
[
{
"node": "LLM Fallback Reply",
"type": "main",
"index": 0
}
]
]
},
"Anthropic Reply": {
"main": [
[
{
"node": "Normalize LLM Output",
"type": "main",
"index": 0
}
],
[
{
"node": "LLM Fallback Reply",
"type": "main",
"index": 0
}
]
]
},
"Normalize LLM Output": {
"main": [
[
{
"node": "Post Digest to Slack",
"type": "main",
"index": 0
}
]
]
},
"LLM Fallback Reply": {
"main": [
[
{
"node": "Post Digest to Slack",
"type": "main",
"index": 0
}
]
]
}
},
"settings": {
"executionOrder": "v1"
}
}
Pro
For the full experience including quality scoring and batch install features for each workflow upgrade to Pro
How this works
Receive a concise daily summary of your team's Slack channel discussions, generated by an advanced language model, to stay informed without sifting through endless messages. This workflow suits busy professionals or managers who rely on Slack for collaboration but need quick insights into key topics, decisions, and action items from the previous day. It fetches the channel's message history via Slack's API, compiles it into a coherent transcript, and routes the analysis to your preferred provider—such as OpenAI or Anthropic—for intelligent summarisation, delivering the digest directly to a designated channel or email.
Use this workflow for high-volume Slack channels where manual review is impractical, like support teams or project groups, ensuring you catch important updates effortlessly. Avoid it for low-activity channels or those requiring real-time alerts, as the daily schedule might miss urgent matters. Common variations include adjusting the time window for weekly digests or integrating additional providers like Google Cloud AI for custom model preferences.
About this workflow
Slack Channel Daily Digest (Multi-Provider LLM). Uses stickyNote, scheduleTrigger, httpRequest, openAi. Scheduled trigger; 14 nodes.
Source: https://github.com/studiomeyer-io/n8n-workflows/blob/main/templates/05-slack-channel-daily-digest/workflow.json — original creator credit. Request a take-down →
Related workflows
Workflows that share integrations, category, or trigger type with this one. All free to copy and import.
AI Institutional Stock Valuation Engine with Risk Scoring & Scenario Targets
Google Sheets, XML, HTTP Request +3
Overview This is a production-grade, fully automated stock analysis system built entirely in n8n. It combines institutional-level financial analysis, dual AI model consensus, and a self-improving back
Google Sheets, XML, HTTP Request +3
This workflow bridges the gap between raw product data and revenue sales tools. It automates the entire Product Qualified Lead (PQL) lifecycle—from real-time intent routing to churn prevention—reducin
HTTP Request, Anthropic, OpenAI
This automation is a complete end-to-end system designed to find, qualify, and contact B2B leads — fully automated and powered by AI. Searches for target companies on LinkedIn via Ghost Genius API, us
Google Sheets, HTTP Request, OpenAI +1
This comprehensive n8n automation template orchestrates a complete end-to-end workflow for generating engaging short-form Point-of-View (POV) style videos using multiple AI services and automatically
HTTP Request, OpenAI, Google Drive +4