This workflow follows the Agent → Chainllm 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": "extract contrat",
"nodes": [
{
"parameters": {},
"type": "n8n-nodes-base.manualTrigger",
"typeVersion": 1,
"position": [
-1020,
540
],
"id": "22f355d0-428a-48d9-9a11-d0b7b7cd2f55",
"name": "When clicking \u2018Test workflow\u2019"
},
{
"parameters": {
"jsCode": "// Debug - Afficher la structure compl\u00e8te des donn\u00e9es re\u00e7ues\nconsole.log('Structure compl\u00e8te des items:', JSON.stringify(items, null, 2));\n\n// R\u00e9cup\u00e9ration des donn\u00e9es binaires du n\u0153ud pr\u00e9c\u00e9dent\nlet base64Data = '';\n\ntry {\n // V\u00e9rifier diff\u00e9rentes structures possibles\n if (items[0].binary) {\n console.log('Donn\u00e9es binaires d\u00e9tect\u00e9es');\n \n // Option 1: Structure Dropbox classique\n if (items[0].binary.data && items[0].binary.data.data) {\n base64Data = items[0].binary.data.data;\n console.log('Donn\u00e9es trouv\u00e9es via binary.data.data');\n }\n // Option 2: Structure formulaire directe\n else if (items[0].binary.data) {\n base64Data = items[0].binary.data;\n console.log('Donn\u00e9es trouv\u00e9es via binary.data');\n }\n // Option 3: Premier champ binaire trouv\u00e9 (formulaire)\n else {\n const binaryKeys = Object.keys(items[0].binary);\n if (binaryKeys.length > 0) {\n const firstBinaryKey = binaryKeys[0];\n if (items[0].binary[firstBinaryKey].data) {\n base64Data = items[0].binary[firstBinaryKey].data;\n console.log(`Donn\u00e9es trouv\u00e9es via binary.${firstBinaryKey}.data`);\n }\n }\n }\n }\n // Option 4: Donn\u00e9es dans JSON\n else if (items[0].json && items[0].json.file) {\n base64Data = items[0].json.file;\n console.log('Donn\u00e9es trouv\u00e9es dans json.file');\n }\n \n // V\u00e9rification finale\n if (!base64Data) {\n console.error('Aucune donn\u00e9e base64 trouv\u00e9e');\n throw new Error('Aucune donn\u00e9e binaire trouv\u00e9e dans la structure');\n }\n \n console.log('Base64 data length:', base64Data.length);\n \n // Retourner la cha\u00eene base64 dans un objet json valide\n return [\n {\n json: {\n data: base64Data,\n debug_info: {\n source: 'binary_extraction',\n data_length: base64Data.length,\n timestamp: new Date().toISOString()\n }\n }\n }\n ];\n \n} catch (error) {\n console.error('Erreur lors de l\\'extraction:', error.message);\n return [\n {\n json: {\n error: error.message,\n debug_structure: items[0]\n }\n }\n ];\n}"
},
"type": "n8n-nodes-base.code",
"typeVersion": 2,
"position": [
-540,
320
],
"id": "b5208812-f6e8-463b-b9b2-f2d621bef219",
"name": "Code"
},
{
"parameters": {
"operation": "download",
"path": "/contrat_ind.pdf"
},
"type": "n8n-nodes-base.dropbox",
"typeVersion": 1,
"position": [
-760,
540
],
"id": "",
"name": "Dropbox",
"credentials": {
"dropboxApi": {
"name": "<your credential>"
}
}
},
{
"parameters": {
"method": "POST",
"url": "https://api.mistral.ai/v1/ocr",
"sendHeaders": true,
"headerParameters": {
"parameters": [
{
"name": "Authorization",
"value": "=Bearer ..."
}
]
},
"sendBody": true,
"specifyBody": "json",
"jsonBody": "={\n \"model\": \"mistral-ocr-latest\",\n \"document\": {\n \"type\": \"document_url\",\n \"document_url\": \"{{ $('Webhook').item.json.body.submission.questions[2].value[0].url }}\"\n },\n \"include_image_base64\": true\n}",
"options": {}
},
"type": "n8n-nodes-base.httpRequest",
"typeVersion": 4.2,
"position": [
-440,
-340
],
"id": "4b313c3b-a17e-40be-90f4-c884b0adb55e",
"name": "HTTP Request"
},
{
"parameters": {
"fieldToSplitOut": "pages",
"options": {}
},
"type": "n8n-nodes-base.splitOut",
"typeVersion": 1,
"position": [
-100,
-340
],
"id": "d8185260-9f2e-419a-aebe-1c21efbf95c1",
"name": "Split Out"
},
{
"parameters": {
"fieldsToSummarize": {
"values": [
{
"aggregation": "concatenate",
"field": "markdown"
}
]
},
"options": {}
},
"type": "n8n-nodes-base.summarize",
"typeVersion": 1.1,
"position": [
260,
-360
],
"id": "df9c9ffa-0a94-4952-8d07-86235b92a3c4",
"name": "Summarize"
},
{
"parameters": {
"promptType": "define",
"text": "=# OBJECTIF\nTu es un assistant charg\u00e9 de transformer un texte issu d'une facture en un email HTML clair et lisible. \nCe texte contient une description (parfois en Markdown) et un bloc JSON structur\u00e9.\n\n# CONSIGNES\n1. Lis et comprends le contenu fourni dans {{ $json.text }}.\n2. Si un bloc JSON est pr\u00e9sent, priorise les donn\u00e9es structur\u00e9es qu\u2019il contient.\n3. G\u00e9n\u00e8re un **r\u00e9sum\u00e9 HTML clair** avec un **design simple et professionnel** :\n - Cr\u00e9e une ou plusieurs sections selon les donn\u00e9es disponibles (ex : client, prestataire, contrat).\n - Chaque section doit appara\u00eetre sous forme de **tableau HTML** bien structur\u00e9.\n - Ne montre que les sections contenant des donn\u00e9es (aucune ligne ou cellule vide).\n4. Applique un **style HTML l\u00e9ger** : bordures fines, titres lisibles, fond clair.\n5. Commence le contenu HTML par une salutation personnalis\u00e9e :\n **\"Bonjour, {{ $('Webhook').item.json.body.submission.questions[0].value }} \"** \n6. Termine le contenu HTML par une formule de politesse :\n **\"Cordialement,<br>HD Automation\"**\n7. Ne fournis que le code HTML propre \u00e0 inclure dans un email Gmail, sans balises `<html>` ou `<head>`, sauf si n\u00e9cessaire pour le rendu.\n\n# STYLE HTML \u00c0 UTILISER\n- Utilise des balises `<table>`, `<th>`, `<td>` pour les sections.\n- Style simple : `border: 1px solid #ddd;`, `padding: 8px;`, `background-color: #f9f9f9;`\n- Un `h2` par section (client, contrat, etc.)\n\n# OBJET D'EMAIL\nEn plus du HTML, g\u00e9n\u00e8re un **objet d\u2019email** clair et professionnel bas\u00e9 sur les infos disponibles :\n- Si possible : `\"R\u00e9sum\u00e9 de votre contrat de service - [Nom Client] - [Mois Ann\u00e9e]\"`\n- Sinon : `\"R\u00e9sum\u00e9 de facture\"`",
"options": {}
},
"type": "@n8n/n8n-nodes-langchain.agent",
"typeVersion": 1.8,
"position": [
960,
-360
],
"id": "4019cea6-2a59-4dae-8754-4727d1d13f30",
"name": "AI Agent"
},
{
"parameters": {
"model": {
"__rl": true,
"value": "gpt-4o",
"mode": "list",
"cachedResultName": "gpt-4o"
},
"options": {}
},
"type": "@n8n/n8n-nodes-langchain.lmChatOpenAi",
"typeVersion": 1.2,
"position": [
1020,
-20
],
"id": "",
"name": "OpenAI Chat Model",
"credentials": {
"openAiApi": {
"name": "<your credential>"
}
}
},
{
"parameters": {
"sendTo": "={{ $('Webhook').item.json.body.submission.questions[1].value }}",
"subject": "={{ /*n8n-auto-generated-fromAI-override*/ $fromAI('Subject', ``, 'string') }}",
"message": "={{ /*n8n-auto-generated-fromAI-override*/ $fromAI('Message', ``, 'string') }}",
"options": {
"appendAttribution": false
}
},
"type": "n8n-nodes-base.gmailTool",
"typeVersion": 2.1,
"position": [
1320,
-20
],
"id": "",
"name": "Gmail",
"credentials": {
"gmailOAuth2": {
"name": "<your credential>"
}
}
},
{
"parameters": {
"formTitle": "contrat",
"formFields": {
"values": [
{
"fieldLabel": "nom client",
"requiredField": true
},
{
"fieldLabel": "contrat",
"fieldType": "file",
"multipleFiles": false,
"acceptFileTypes": ".pdf",
"requiredField": true
},
{
"fieldLabel": "email",
"fieldType": "email",
"requiredField": true
}
]
},
"options": {}
},
"type": "n8n-nodes-base.formTrigger",
"typeVersion": 2.2,
"position": [
-920,
340
],
"id": "c3560b69-cee5-495e-8e9d-1920c0a5ef95",
"name": "On form submission"
},
{
"parameters": {
"jsCode": "// R\u00e9cup\u00e9ration des donn\u00e9es binaires du n\u0153ud pr\u00e9c\u00e9dent\nconst binaryData = items[0].binary.data;\nlet base64Data = '';\n\nif (binaryData) {\n // Extraction du base64 seulement\n base64Data = binaryData.data;\n \n // Retourner la cha\u00eene base64 dans un objet json valide\n items = [\n {\n json: {\n data: base64Data\n }\n }\n ];\n}\n\nreturn items;"
},
"type": "n8n-nodes-base.code",
"typeVersion": 2,
"position": [
-460,
540
],
"id": "f11079e2-615f-465c-81e7-b8d567e3e654",
"name": "Code1"
},
{
"parameters": {
"promptType": "define",
"text": "=peut tu ma structurer cette donn\u00e9e {{ $json.concatenated_markdown }}",
"hasOutputParser": true,
"messages": {
"messageValues": [
{
"message": "=#OBJECTIF \nTu es un assistant qui structure les donn\u00e9es issues d'une facture afin de les rendre lisibles pour un client et exploitables automatiquement.\n\n#CONSIGNES \nTu re\u00e7ois un texte brut extrait d\u2019une facture (OCR ou formulaire). \nTu dois me fournir **deux r\u00e9sultats distincts** :\n\n---\n\n## 1. \u2705 R\u00e9sum\u00e9 Markdown (lisible humain) \nOrganis\u00e9 clairement avec des titres et tableaux en Markdown. Doit inclure :\n- Informations du **client**\n- Informations de la **soci\u00e9t\u00e9 \u00e9mettrice**\n- D\u00e9tails de la **facture** (num\u00e9ro, date, conditions\u2026)\n- D\u00e9tails des **prestations / produits**\n- Totaux (HT, TVA, TTC)\n\n---\n\n## 2. \u2705 Bloc JSON clair et structur\u00e9 (machine-friendly) \nTu dois :\n- Structurer les donn\u00e9es en **objets simples** (`client`, `societe`, `facture`, `totaux`)\n- **Supprimer tous les champs vides ou non renseign\u00e9s**\n- N'utiliser un **tableau uniquement pour les prestations**"
}
]
}
},
"type": "@n8n/n8n-nodes-langchain.chainLlm",
"typeVersion": 1.5,
"position": [
540,
-360
],
"id": "38c5b03d-7701-4371-9187-33529563b777",
"name": "Basic LLM Chain"
},
{
"parameters": {
"model": {
"__rl": true,
"value": "gpt-4o",
"mode": "list",
"cachedResultName": "gpt-4o"
},
"options": {}
},
"type": "@n8n/n8n-nodes-langchain.lmChatOpenAi",
"typeVersion": 1.2,
"position": [
540,
-100
],
"id": "",
"name": "OpenAI Chat Model1",
"credentials": {
"openAiApi": {
"name": "<your credential>"
}
}
},
{
"parameters": {
"httpMethod": "POST",
"path": "8b34fed3-1ff6-4884-bd0a-a729617c6db2",
"options": {}
},
"type": "n8n-nodes-base.webhook",
"typeVersion": 2,
"position": [
-820,
-320
],
"id": "35928664-70b5-4001-8963-01c6634c401b",
"name": "Webhook"
},
{
"parameters": {
"jsCode": "// Debug - Afficher la structure compl\u00e8te des donn\u00e9es re\u00e7ues\nconsole.log('Structure compl\u00e8te des items:', JSON.stringify(items, null, 2));\n\nlet fileUrl = '';\ntry {\n // Cas : URL du fichier dans items[0].json.fileUrl\n if (items[0].json && items[0].json.fileUrl) {\n fileUrl = items[0].json.fileUrl;\n console.log('URL trouv\u00e9e dans json.fileUrl');\n } \n // Cas alternatif : URL dans un autre champ JSON\n else if (items[0].json) {\n const keys = Object.keys(items[0].json);\n for (const key of keys) {\n if (typeof items[0].json[key] === 'string' && items[0].json[key].startsWith('http')) {\n fileUrl = items[0].json[key];\n console.log(`URL trouv\u00e9e dans json.${key}`);\n break;\n }\n }\n }\n\n if (!fileUrl) {\n throw new Error('Aucune URL de fichier trouv\u00e9e dans la structure');\n }\n\n // Retourne l'URL du fichier\n return [\n {\n json: {\n fileUrl,\n debug_info: {\n source: 'url_extraction',\n timestamp: new Date().toISOString()\n }\n }\n }\n ];\n\n} catch (error) {\n console.error('Erreur lors de l\\'extraction:', error.message);\n return [\n {\n json: {\n error: error.message,\n debug_structure: items[0]\n }\n }\n ];\n}\n"
},
"type": "n8n-nodes-base.code",
"typeVersion": 2,
"position": [
-1200,
840
],
"id": "a8f57e83-4aed-499d-8d30-d5ae56b02491",
"name": "Code2"
},
{
"parameters": {
"promptType": "define",
"text": "=# OBJECTIF\nTu es un assistant charg\u00e9 de transformer un texte issu d'une facture en un email clair et lisible. \nCe texte contient une description (souvent en Markdown) et un bloc JSON. \nTu dois produire un **r\u00e9sum\u00e9 HTML lisible sous forme de tableaux bien structur\u00e9s** et l\u2019envoyer par **email via Gmail**.\n\n# CONSIGNES\n\n1. Lis et comprends le contenu de {{ $json.text }}.\n2. Si un bloc JSON est pr\u00e9sent, priorise les donn\u00e9es structur\u00e9es qu'il contient.\n3. G\u00e9n\u00e8re un **r\u00e9sum\u00e9 HTML clair** avec un **design simple et professionnel** :\n - Une ou plusieurs sections (client, prestataire, contrat, etc.) selon les informations disponibles.\n - Chaque section dans un tableau HTML.\n - Ne g\u00e9n\u00e8re que les sections disponibles (aucune valeur vide).\n4. Ajoute un **style HTML de base** (bordures fines, fond clair, titres lisibles).\n5. R\u00e9dige un **objet d\u2019email clair**, bas\u00e9 sur les informations disponibles. Par exemple :\n - *\"R\u00e9sum\u00e9 de votre contrat de service - Marie Dupont - Avril 2025\"*\n - Si nom/date non dispo, utilise un titre g\u00e9n\u00e9rique : *\"R\u00e9sum\u00e9 de facture \"*.\n6. Le contenu HTML sera utilis\u00e9 comme corps d'email dans un envoi via Gmail.\n\n# STYLE HTML \u00c0 UTILISER\nhtml\n<style>\n table {\n border-collapse: collapse;\n width: 100%;\n margin: 20px 0;\n font-family: Arial, sans-serif;\n }\n th, td {\n border: 1px solid #ddd;\n padding: 8px 12px;\n }\n th {\n background-color: #f9f9f9;\n font-weight: bold;\n }\n h2 {\n font-family: Arial, sans-serif;\n color: #444;\n margin-top: 30px;\n }\n</style>\n7. Le message HTML doit commencer par une salutation personnalis\u00e9e :\n - **\"Bonjour, {{ $('Webhook').item.json.body.submission.questions[0].value }} \"**\n \n\n8. \u00c0 la fin du contenu HTML, ajoute une formule de politesse :\n - **\"Cordialement,<br> HD Automation\"**\n",
"options": {}
},
"type": "@n8n/n8n-nodes-langchain.agent",
"typeVersion": 1.8,
"position": [
-800,
720
],
"id": "5eb6d1ff-a023-4b73-9e57-3f65b15063ae",
"name": "AI Agent1"
},
{
"parameters": {
"model": {
"__rl": true,
"value": "gpt-4o",
"mode": "list",
"cachedResultName": "gpt-4o"
},
"options": {}
},
"type": "@n8n/n8n-nodes-langchain.lmChatOpenAi",
"typeVersion": 1.2,
"position": [
-780,
1020
],
"id": "",
"name": "OpenAI Chat Model2",
"credentials": {
"openAiApi": {
"name": "<your credential>"
}
}
},
{
"parameters": {
"sendTo": "={{ $('Webhook').item.json.body.submission.questions[1].value }}",
"subject": "={{ /*n8n-auto-generated-fromAI-override*/ $fromAI('Subject', ``, 'string') }}",
"message": "={{ /*n8n-auto-generated-fromAI-override*/ $fromAI('Message', ``, 'string') }}",
"options": {
"appendAttribution": false
}
},
"type": "n8n-nodes-base.gmailTool",
"typeVersion": 2.1,
"position": [
-580,
960
],
"id": "598fa449-d89e-44a5-a67f-18b2af8757a9",
"name": "Gmail1",
"credentials": {
"gmailOAuth2": {
"name": "<your credential>"
}
}
},
{
"parameters": {
"sessionIdType": "customKey",
"sessionKey": "={{ $('Webhook').item.json.body.submission.questions[0].value }}",
"contextWindowLength": 10000
},
"type": "@n8n/n8n-nodes-langchain.memoryBufferWindow",
"typeVersion": 1.3,
"position": [
1140,
0
],
"id": "aff273bf-9c03-45c5-aa12-43aa5a9aa6e0",
"name": "Simple Memory"
}
],
"connections": {
"When clicking \u2018Test workflow\u2019": {
"main": [
[
{
"node": "Dropbox",
"type": "main",
"index": 0
}
]
]
},
"Dropbox": {
"main": [
[
{
"node": "Code1",
"type": "main",
"index": 0
}
]
]
},
"Code": {
"main": [
[]
]
},
"HTTP Request": {
"main": [
[
{
"node": "Split Out",
"type": "main",
"index": 0
}
]
]
},
"Split Out": {
"main": [
[
{
"node": "Summarize",
"type": "main",
"index": 0
}
]
]
},
"Summarize": {
"main": [
[
{
"node": "Basic LLM Chain",
"type": "main",
"index": 0
}
]
]
},
"OpenAI Chat Model": {
"ai_languageModel": [
[
{
"node": "AI Agent",
"type": "ai_languageModel",
"index": 0
}
]
]
},
"Gmail": {
"ai_tool": [
[
{
"node": "AI Agent",
"type": "ai_tool",
"index": 0
}
]
]
},
"On form submission": {
"main": [
[
{
"node": "Code",
"type": "main",
"index": 0
}
]
]
},
"OpenAI Chat Model1": {
"ai_languageModel": [
[
{
"node": "Basic LLM Chain",
"type": "ai_languageModel",
"index": 0
}
]
]
},
"Basic LLM Chain": {
"main": [
[
{
"node": "AI Agent",
"type": "main",
"index": 0
}
]
]
},
"Webhook": {
"main": [
[
{
"node": "HTTP Request",
"type": "main",
"index": 0
}
]
]
},
"Code2": {
"main": [
[]
]
},
"OpenAI Chat Model2": {
"ai_languageModel": [
[
{
"node": "AI Agent1",
"type": "ai_languageModel",
"index": 0
}
]
]
},
"Gmail1": {
"ai_tool": [
[
{
"node": "AI Agent1",
"type": "ai_tool",
"index": 0
}
]
]
},
"Simple Memory": {
"ai_memory": [
[
{
"node": "AI Agent",
"type": "ai_memory",
"index": 0
}
]
]
}
},
"active": true,
"settings": {
"executionOrder": "v1"
},
"versionId": "3eb09777-2136-4dde-bb3b-5a39bdb05527",
"meta": {
"templateCredsSetupCompleted": true
},
"id": "QxMWqc2ZnyUOPyhc",
"tags": []
}
Credentials you'll need
Each integration node will prompt for credentials when you import. We strip credential IDs before publishing — you'll add your own.
dropboxApigmailOAuth2openAiApi
Pro
For the full experience including quality scoring and batch install features for each workflow upgrade to Pro
About this workflow
extract contrat. Uses dropbox, httpRequest, agent, lmChatOpenAi. Event-driven trigger; 19 nodes.
Source: https://github.com/YassineElabdaoui/AgentIA-Facture/blob/fa570a64c208d604df3a71c121da75785cca604e/extract_contrat.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.
Inbox Guardian. Uses gmailTrigger, lmChatOpenAi, agent, textClassifier. Event-driven trigger; 66 nodes.
Gmail Trigger, OpenAI Chat, Agent +12
This workflow serves as a comprehensive "Workflow Nodes SEO & Documentation Generator". It uses AI to analyze, rename, and document n8n workflows, offering a streamlined way to optimize workflow reada
Form Trigger, n8n, Output Parser Autofixing +11
48_WAgentEnhancement. Uses whatsAppTrigger, whatsApp, openAi, httpRequest. Event-driven trigger; 56 nodes.
WhatsApp Trigger, WhatsApp, OpenAI +13
PixelSensei(ZH). Uses agent, outputParserStructured, formTrigger, lmChatOpenAi. Event-driven trigger; 55 nodes.
Agent, Output Parser Structured, Form Trigger +4
Lection 9 main. Uses formTrigger, chatTrigger, agent, lmChatOpenAi. Event-driven trigger; 55 nodes.
Form Trigger, Chat Trigger, Agent +7