Auto-Generate a Visual Daily Journal from Discord Chats to Notion

Input: Discord channel messages → Process: scheduled trigger, filter today’s messages, GPT-4 summary & image prompt, DALL-E 3 image, Cloudinary hosting → Output: Notion diary page with title, summary, mood, tags, and AI cover image

Who

Individuals or small teams who communicate daily in Discord and want an automated diary

You get

Every day, fetch messages from a Discord channel, use GPT-4 to generate a diary title, summary, mood, and tags, generate a vibe-matching image with DALL-E 3, and save everything into a Notion database with Gallery view

You need

n8n · discord · cloudinary · notion

Steps

1. Configure schedule trigger and Discord message fetch

   Add a Schedule Trigger node and set your daily run time (default 11 PM). Then add a Discord node (resource=message, operation=getAll) and select the target server and channel. You must configure a Discord Bot API credential in n8n Settings → Credentials, invite the bot to the server, and grant it Read Message History permission.

2. Filter today’s messages and check for non-empty results

   Add a Code node named 'Filter Today's Messages' to keep only messages between 00:00 and 23:59 of the current day, formatting them into an array with id, content, author, timestamp, and attachments (chronological order). Then add an If node named 'Has Messages?' that checks whether messageCount > 0. If zero, the workflow branches to 'No Messages Today' and skips the diary entry.

   Execution prompt · Source text。Code node 'Filter Today's Messages' JS Code field

   Copy prompt

   // Filter messages from today only
   const allMessages = $input.all();
   const now = new Date();
   const todayStart = new Date(now.getFullYear(), now.getMonth(), now.getDate(), 0, 0, 0);
   const todayEnd = new Date(now.getFullYear(), now.getMonth(), now.getDate(), 23, 59, 59);
   
   // Filter to today's messages
   const todayMessages = allMessages.filter(item => {
     const msgDate = new Date(item.json.timestamp);
     return msgDate >= todayStart && msgDate <= todayEnd;
   });
   
   // Format the messages
   const formattedMessages = todayMessages.map(item => ({
     id: item.json.id,
     content: item.json.content || '',
     author: item.json.author?.username || 'Unknown',
     timestamp: item.json.timestamp,
     attachments: item.json.attachments?.length || 0
   })).reverse(); // Chronological order
   
   return {
     json: {
       messages: formattedMessages,
       messageCount: formattedMessages.length,
       date: now.toISOString().split('T')[0],
       fetchedAt: now.toISOString()
     }
   };

3. Format conversation and call GPT-4 to generate diary content

   Add a Code node named 'Format for GPT-4' that concatenates filtered messages into a timestamped conversation text, extracts unique participants and per-person message counts, and produces a formatted date string. Then add an HTTP Request node named 'GPT-4 Analyze Day' that POSTs to the OpenAI Chat Completions API (model=gpt-4o) with a system prompt and user prompt containing the conversation, enforcing a JSON Schema response for title, summary, mood, tags, imagePrompt, and highlights. Configure the OpenAI API credential in n8n Settings → Credentials.

   Execution prompt · Source text。Code node 'Format for GPT-4' JS Code field

   Copy prompt

   // Format messages for GPT-4 analysis
   const data = $input.first().json;
   const messages = data.messages;
   
   if (!messages || messages.length === 0) {
     return { json: { error: 'No messages to process' } };
   }
   
   // Build conversation text
   const conversationText = messages.map(msg => {
     const time = new Date(msg.timestamp).toLocaleTimeString('en-US', {
       hour: '2-digit',
       minute: '2-digit'
     });
     return `[${time}] ${msg.author}: ${msg.content}`;
   }).join('\n');
   
   // Extract unique participants
   const participants = [...new Set(messages.map(m => m.author))];
   
   // Count messages per person
   const messageCounts = {};
   messages.forEach(msg => {
     messageCounts[msg.author] = (messageCounts[msg.author] || 0) + 1;
   });
   
   return {
     json: {
       conversationText: conversationText,
       participants: participants,
       messageCounts: messageCounts,
       totalMessages: data.messageCount,
       date: data.date,
       dateFormatted: new Date(data.date).toLocaleDateString('en-US', {
         weekday: 'long',
         year: 'numeric',
         month: 'long',
         day: 'numeric'
       })
     }
   };

   Execution prompt · Source text。HTTP Request node 'GPT-4 Analyze Day' JSON Body field (n8n expression syntax preserved)

   Copy prompt

   ={
     "model": "gpt-4o",
     "messages": [
       {
         "role": "system",
         "content": "You are a thoughtful journal assistant. Analyze the day's Discord conversations and create a personal diary entry. Be warm, reflective, and capture the essence of the day. Focus on themes, emotions, achievements, and memorable moments."
       },
       {
         "role": "user",
         "content": "Here are today's Discord messages ({{ $json.dateFormatted }}):\n\n{{ $json.conversationText }}\n\nParticipants: {{ $json.participants.join(', ') }}\nTotal messages: {{ $json.totalMessages }}\n\nPlease analyze this and provide:\n1. A diary entry title (creative, captures the day's essence)\n2. A reflective summary (2-3 paragraphs, personal tone)\n3. Mood assessment (one of: Great, Good, Neutral, Low, Productive)\n4. 3-5 relevant tags (topics discussed)\n5. An image prompt to visually represent this day (artistic, atmospheric, no text)"
       }
     ],
     "response_format": {
       "type": "json_schema",
       "json_schema": {
         "name": "diary_entry",
         "strict": true,
         "schema": {
           "type": "object",
           "additionalProperties": false,
           "properties": {
             "title": { "type": "string" },
             "summary": { "type": "string" },
             "mood": { "type": "string" },
             "tags": {
               "type": "array",
               "items": { "type": "string" }
             },
             "imagePrompt": { "type": "string" },
             "highlights": {
               "type": "array",
               "items": { "type": "string" }
             }
           },
           "required": ["title", "summary", "mood", "tags", "imagePrompt", "highlights"]
         }
       }
     }
   }

4. Parse GPT-4 output and generate DALL-E cover image

   Add a Code node named 'Parse Analysis' that parses the GPT-4 JSON response and maps the mood to an emoji (Great=😊, Good=😌, Neutral=😐, Low=😔, Productive=🔥). Then add a Code node named 'Generate Image (DALL-E)' that calls the OpenAI Images Generations API (model=dall-e-3, size=1024x1024, response_format=b64_json) via fetch, appending an artistic style suffix to the original imagePrompt. Important: this node contains a hard-coded placeholder OPENAI_API_KEY = 'YOUR_OPENAI_API_KEY'; replace it with a real key or migrate to n8n Credentials before production use.

   Execution prompt · Source text。Code node 'Parse Analysis' JS Code field

   Copy prompt

   // Parse GPT-4 response and prepare for image generation
   const response = $input.first().json;
   const previousData = $('Format for GPT-4').first().json;
   
   const analysis = JSON.parse(response.choices[0].message.content);
   
   // Map mood to emoji
   const moodEmojis = {
     'Great': '😊',
     'Good': '😌',
     'Neutral': '😐',
     'Low': '😔',
     'Productive': '🔥'
   };
   
   const moodWithEmoji = `${moodEmojis[analysis.mood] || '😐'} ${analysis.mood}`;
   
   return {
     json: {
       title: analysis.title,
       summary: analysis.summary,
       mood: analysis.mood,
       moodWithEmoji: moodWithEmoji,
       tags: analysis.tags,
       imagePrompt: analysis.imagePrompt,
       highlights: analysis.highlights,
       date: previousData.date,
       dateFormatted: previousData.dateFormatted,
       totalMessages: previousData.totalMessages,
       participants: previousData.participants
     }
   };

   Execution prompt · Source text。Code node 'Generate Image (DALL-E)' JS Code field

   Copy prompt

   // ============================================
   // GENERATE IMAGE WITH DALL-E 3
   // ============================================
   // ⚠️ UPDATE THIS API KEY
   const OPENAI_API_KEY = 'YOUR_OPENAI_API_KEY';
   // ============================================
   
   const data = $input.first().json;
   
   // Enhance the prompt for better artistic results
   const enhancedPrompt = `${data.imagePrompt}. Artistic digital illustration, dreamy atmosphere, soft lighting, no text or words, suitable for a personal journal.`;
   
   try {
     const response = await fetch('https://api.openai.com/v1/images/generations', {
       method: 'POST',
       headers: {
         'Authorization': `Bearer ${OPENAI_API_KEY}`,
         'Content-Type': 'application/json'
       },
       body: JSON.stringify({
         model: 'dall-e-3',
         prompt: enhancedPrompt.substring(0, 4000),
         n: 1,
         size: '1024x1024',
         quality: 'standard',
         response_format: 'b64_json'
       })
     });
     
     const result = await response.json();
     
     if (result.error) {
       return {
         json: {
           ...data,
           imageBase64: null,
           imageError: result.error.message
         }
       };
     }
     
     return {
       json: {
         ...data,
         imageBase64: result.data[0].b64_json,
         imageError: null
       }
     };
     
   } catch (error) {
     return {
       json: {
         ...data,
         imageBase64: null,
         imageError: error.message
       }
     };
   }

5. Upload image to Cloudinary and write to Notion

   Add a Code node named 'Upload to Cloudinary' that uploads the base64 image using a SHA1-signed request to Cloudinary (folder='visual-journal'), returning a public secure_url. Then add a Notion node named 'Create Notion Entry' that creates a new page in your database, populating Title, Date, Summary, Mood (with emoji), and Message Count properties, and inserting page blocks for the image, a Summary heading, a Highlights bulleted list, a divider, and participant/message-count stats. Optionally add a 'Success Response' Code node to output a summary of the created entry.

   Execution prompt · Source text。Code node 'Upload to Cloudinary' JS Code field

   Copy prompt

   // ============================================
   // UPLOAD IMAGE TO CLOUDINARY
   // ============================================
   // ⚠️ UPDATE THESE VALUES
   const CLOUDINARY_CLOUD_NAME = 'YOUR_CLOUD_NAME';
   const CLOUDINARY_API_KEY = 'YOUR_API_KEY';
   const CLOUDINARY_API_SECRET = 'YOUR_API_SECRET';
   // ============================================
   
   const data = $input.first().json;
   
   if (!data.imageBase64) {
     return {
       json: {
         ...data,
         imageUrl: null,
         uploadError: data.imageError || 'No image to upload'
       }
     };
   }
   
   try {
     // Create signature for Cloudinary
     const timestamp = Math.floor(Date.now() / 1000);
     const folder = 'visual-journal';
     const publicId = `diary_${data.date}`;
     
     // Build signature string
     const signatureString = `folder=${folder}&public_id=${publicId}&timestamp=${timestamp}${CLOUDINARY_API_SECRET}`;
     
     // Create SHA1 hash
     const crypto = require('crypto');
     const signature = crypto.createHash('sha1').update(signatureString).digest('hex');
     
     // Upload to Cloudinary
     const formData = new URLSearchParams();
     formData.append('file', `data:image/png;base64,${data.imageBase64}`);
     formData.append('api_key', CLOUDINARY_API_KEY);
     formData.append('timestamp', timestamp.toString());
     formData.append('signature', signature);
     formData.append('folder', folder);
     formData.append('public_id', publicId);
     
     const response = await fetch(
       `https://api.cloudinary.com/v1_1/${CLOUDINARY_CLOUD_NAME}/image/upload`,
       {
         method: 'POST',
         body: formData
       }
     );
     
     const result = await response.json();
     
     if (result.error) {
       return {
         json: {
           ...data,
           imageUrl: null,
           uploadError: result.error.message
         }
       };
     }
     
     return {
       json: {
         ...data,
         imageUrl: result.secure_url,
         imagePublicId: result.public_id,
         uploadError: null
       }
     };
     
   } catch (error) {
     return {
       json: {
         ...data,
         imageUrl: null,
         uploadError: error.message
       }
     };
   }

6. (Optional) Skip when no messages found

   If the If node determines messageCount = 0, the workflow branches to 'No Messages Today', returning a skip message without creating a Notion page.

   Execution prompt · Source text。Code node 'No Messages Today' JS Code field

   Copy prompt

   // No messages today - skip entry
   const data = $input.first().json;
   
   return {
     json: {
       success: false,
       message: 'No messages found for today. Skipping diary entry.',
       date: data.date || new Date().toISOString().split('T')[0]
     }
   };

You are an n8n workflow setup assistant. The user wants to recreate the 'Discord Daily Visual Journal to Notion' template. First confirm they have: an n8n instance, a Discord Bot with Read Message History permission, an OpenAI API key, a Cloudinary account, and a Notion integration with a properly structured database. Then guide them through importing the template or building manually, checking: 1) Discord credential and channel selection; 2) OpenAI credential setup (HTTP Request node uses predefinedCredentialType=openAiApi); 3) Replacing the OPENAI_API_KEY placeholder in the DALL-E Code node; 4) Replacing Cloudinary Cloud Name/API Key/Secret in the Upload node; 5) Notion databaseId and sharing the page with the integration. Remind them: GPT-4o and DALL-E 3 incur API costs; Cloudinary free tier is usually enough; hard-coded keys in nodes should ideally be moved to n8n Credentials.