Générer des publications sociales Twitter et LinkedIn à partir de poussées GitHub
Ceci est unMultimodal AIworkflow d'automatisation du domainecontenant 18 nœuds.Utilise principalement des nœuds comme If, Merge, Github, Twitter, Webhook. Génération automatique de publications pour les médias sociaux à partir de mises à jour GitHub README/CHANGELOG avec GPT-4o et OAuth
- •Personal Access Token GitHub
- •Informations d'identification Twitter API
- •Point de terminaison HTTP Webhook (généré automatiquement par n8n)
- •Informations d'identification LinkedIn API
- •Clé API OpenAI
Nœuds utilisés (18)
Catégorie
{
"id": "zMl1xguA65ATPxMh",
"meta": {
"instanceId": "6bad1eba185697a960de5bc9203d6bff2fd56440a0b5303c382e0b1537a2a568",
"templateCredsSetupCompleted": true
},
"name": "Generate social posts from GitHub pushes to Twitter and LinkedIn",
"tags": [],
"nodes": [
{
"id": "29f468dd-8822-4de7-9bf7-0205e8b51b12",
"name": "Webhook",
"type": "n8n-nodes-base.webhook",
"position": [
-640,
-144
],
"webhookId": "70b80513-b93c-4749-9a06-e78d4f1f2d23",
"parameters": {
"path": "70b80513-b93c-4749-9a06-e78d4f1f2d23",
"options": {},
"httpMethod": "POST"
},
"typeVersion": 2
},
{
"id": "a991ef0a-48ef-4ccf-98c6-587edddafa96",
"name": "Fusionner",
"type": "n8n-nodes-base.merge",
"position": [
400,
-80
],
"parameters": {},
"typeVersion": 3.2
},
{
"id": "a19b66f5-0927-4c59-8343-f2cea29feb5e",
"name": "Obtenir README",
"type": "n8n-nodes-base.github",
"position": [
64,
-208
],
"webhookId": "eab25bcf-6820-4ee6-ab43-7ec9f5a9e413",
"parameters": {
"owner": {
"__rl": true,
"mode": "name",
"value": "jorge210488"
},
"filePath": "README.md",
"resource": "file",
"operation": "get",
"repository": {
"__rl": true,
"mode": "name",
"value": "={{ $json.body.repository.name }}"
},
"additionalParameters": {}
},
"credentials": {
"githubApi": {
"id": "O2s7MSmQAwSFAbN5",
"name": "GitHub account"
}
},
"typeVersion": 1.1
},
{
"id": "b3ae8e15-dd37-4207-a78c-4b18b72c9cb8",
"name": "Obtenir CHANGELOG",
"type": "n8n-nodes-base.github",
"position": [
64,
32
],
"webhookId": "eab25bcf-6820-4ee6-ab43-7ec9f5a9e413",
"parameters": {
"owner": {
"__rl": true,
"mode": "name",
"value": "jorge210488"
},
"filePath": "CHANGELOG.md",
"resource": "file",
"operation": "get",
"repository": {
"__rl": true,
"mode": "name",
"value": "={{ $json.body.repository.name }}"
},
"additionalParameters": {}
},
"credentials": {
"githubApi": {
"id": "O2s7MSmQAwSFAbN5",
"name": "GitHub account"
}
},
"typeVersion": 1.1
},
{
"id": "9229a69b-fe35-4a25-a8d8-bc3e1f6d8894",
"name": "Si",
"type": "n8n-nodes-base.if",
"position": [
-448,
-144
],
"parameters": {
"options": {},
"conditions": {
"options": {
"version": 2,
"leftValue": "",
"caseSensitive": true,
"typeValidation": "strict"
},
"combinator": "and",
"conditions": [
{
"id": "fb7ff717-d8a0-44ed-9053-1e88d50f88de",
"operator": {
"type": "boolean",
"operation": "true",
"singleValue": true
},
"leftValue": "={{\n ($json.body.commits ?? []).some(c =>\n [ ...(c.added ?? []), ...(c.modified ?? []), ...(c.removed ?? []) ]\n .some(f => String(f).toLowerCase().includes('changelog'))\n )\n}}",
"rightValue": "="
}
]
}
},
"typeVersion": 2.2
},
{
"id": "4e9ebbbd-9c37-4dbe-a9b6-dc0f72a85877",
"name": "Aucune opération, ne rien faire",
"type": "n8n-nodes-base.noOp",
"position": [
-256,
48
],
"parameters": {},
"typeVersion": 1
},
{
"id": "e35435d4-f3f8-44f1-ac78-21efda7076ac",
"name": "Agréger",
"type": "n8n-nodes-base.aggregate",
"position": [
560,
-80
],
"parameters": {
"options": {},
"aggregate": "aggregateAllItemData"
},
"typeVersion": 1
},
{
"id": "82c95a95-4a00-405d-a826-9c387a3f76be",
"name": "Chaîne LLM de base",
"type": "@n8n/n8n-nodes-langchain.chainLlm",
"position": [
832,
-80
],
"parameters": {
"text": "=README: {{ $json.data[0].data }}\nCHANGELOG: {{ $json.data[1].data }}",
"batching": {},
"messages": {
"messageValues": [
{
"message": "=# Role\n\nYou are a precise content generator for social media.\nYour input will be a JSON object with two string fields:\n\n* `README`: full project README in Markdown.\n* `CHANGELOG`: full changelog in Markdown.\n\n## Your task\n\nProduce **one JSON object** with exactly two keys:\n\n```json\n{\n \"twitter\": \"<tweet in English, <=280 chars including the GitHub link>\",\n \"linkedin\": \"<LinkedIn post in English, long-form>\"\n}\n```\n\nNo extra keys, no surrounding text, no Markdown fences.\n\n## Parsing rules (do not hallucinate)\n\n1. **Repository URL (required in both posts):**\n\n * Extract the **first** URL that matches a GitHub repo pattern from `CHANGELOG` using a regex like:\n `https?://github\\.com/[A-Za-z0-9_.-]+/[A-Za-z0-9_.-]+`\n * If not found in `CHANGELOG`, search `README` with the same regex.\n * If still not found, **do not fabricate** a URL; instead, use the literal placeholder `https://github.com/REPO_MISSING`. (Keep generating the posts.)\n\n2. **Project name:**\n\n * Prefer the first H1 (`# Title`) in `README`. If not present, infer a concise name from the first heading found. If impossible, use a short neutral name like “This project”.\n\n3. **First release or update determination:**\n\n * From `CHANGELOG`, detect if this is the **first or only version** (e.g., version numbers like `v1.0.0`, \"Initial release\", or only one version section present).\n * If first/only version → the post should clearly state it's a **new project/app/development** (e.g., “New project”, “Launching a new application”).\n * If not first version → the post should state it’s an **update or improvement** (e.g., “Updating the project X”, “Improving the application X”).\n\n4. **Description & features:**\n\n * Derive a clear, one-sentence value proposition from `README`’s Description/Overview.\n * Collect 4–7 key capabilities/benefits (bullets for LinkedIn).\n\n5. **Tech stack:**\n\n * From `README`, extract notable technologies (backend, frontend, infra/services).\n * Examples to detect if present: Django, DRF, Celery, Redis, Channels/WebSockets, Next.js, React, TypeScript, Tailwind CSS, Zustand, Framer Motion, Stripe, OpenAI, OAuth/Google, SMTP/Email, Docker, Docker Compose.\n * For **Twitter**, mention only the **3–4 most central** technologies.\n * For **LinkedIn**, you may group by Backend / Frontend / Infra.\n\n## LinkedIn post (English, long-form)\n\n* Tone: professional, enthusiastic, concise sentences; **no emojis**.\n* Structure (use short paragraphs and bullet points):\n\n * Hook:\n\n * If first release → start with “New project: …” or similar phrasing.\n * If update → start with “Updating the project \\[name]: …” or similar phrasing.\n * What it does: 1 short paragraph.\n * Key features/benefits: 4–7 bullets.\n * Tech stack (grouped): bullets for Backend, Frontend, Infra/Services.\n * Motivational developer line (one sentence, non-cheesy).\n * Call to action with the **repo link** (exactly once).\n* Hashtags: **8–12** relevant tags, all lowercase, no spaces (use hyphens only if part of the tech brand). Prioritize technologies and domain, e.g.:\n `#python #django #nextjs #typescript #tailwindcss #stripe #openai #celery #redis #websockets #ai #saas #startups #webdevelopment`\n (Pick only those truly present; don’t invent.)\n\n## Twitter post (English, max 280 chars including link)\n\n* Format:\n\n * If first release → short sentence like “New project: …” or similar.\n * If update → short sentence like “Updating project \\[name]: …” or similar.\n* Describe what the app does + **repo link** (once).\n* Mention **3–4** main technologies inline (e.g., Django, Next.js, Stripe, OpenAI).\n* **2–3 hashtags max**, chosen from the most relevant.\n* **No emojis.**\n* Length enforcement: if >280 chars, iteratively shorten by (in order): remove adjectives, compress wording, then drop one hashtag at a time until ≤280. Never drop the repo link.\n\n## Output formatting\n\n* Return a **single JSON object** with keys `\"twitter\"` and `\"linkedin\"`.\n* Escape quotes properly.\n* Do **not** add explanations, markdown, or extra whitespace before/after the JSON.\n* Do not include any links other than the repo link.\n\n## Safety & accuracy\n\n* Never fabricate tech or features not present in the inputs.\n* If some sections are missing in inputs, omit that detail gracefully.\n* Keep everything in **English** except hashtags which are typically lowercase English tokens.\n* Do not output dates/releases unless explicitly present.\n* Maintain a positive, credible tone.\n"
}
]
},
"promptType": "define",
"hasOutputParser": true
},
"typeVersion": 1.7
},
{
"id": "927afa09-65bf-4c62-955c-42a3bb098f28",
"name": "OpenAI Modèle de chat",
"type": "@n8n/n8n-nodes-langchain.lmChatOpenAi",
"position": [
832,
80
],
"parameters": {
"model": {
"__rl": true,
"mode": "list",
"value": "gpt-4o-mini"
},
"options": {}
},
"credentials": {
"openAiApi": {
"id": "9RjbvMHnwZg2LraO",
"name": "OpenAi ROD"
}
},
"typeVersion": 1.2
},
{
"id": "43f2f142-8319-445f-80a9-61c56ae6ddad",
"name": "Analyseur de sortie structurée",
"type": "@n8n/n8n-nodes-langchain.outputParserStructured",
"position": [
1008,
80
],
"parameters": {
"jsonSchemaExample": "{\n \"twitter\": \"tweet in English, <=280 chars including the GitHub link\",\n \"linkedin\": \"LinkedIn post in English, long-form\"\n}"
},
"typeVersion": 1.2
},
{
"id": "c2454817-d07d-45e0-8c5f-03a68c619eb8",
"name": "Publier un tweet",
"type": "n8n-nodes-base.twitter",
"position": [
1248,
48
],
"parameters": {
"text": "={{ $json.output.twitter }}",
"additionalFields": {}
},
"credentials": {
"twitterOAuth2Api": {
"id": "VF7lxb5r1ouC0O22",
"name": "X account"
}
},
"typeVersion": 2
},
{
"id": "944f1559-f79e-4034-b70f-0246c8b40a5f",
"name": "LinkedIn",
"type": "n8n-nodes-base.linkedIn",
"position": [
1248,
-160
],
"parameters": {
"text": "={{ $json.output.linkedin }}",
"person": "nUdV-_cHkk",
"additionalFields": {}
},
"credentials": {
"linkedInOAuth2Api": {
"id": "Slo13uTVYAdaV30t",
"name": "LinkedIn account"
}
},
"typeVersion": 1
},
{
"id": "d64cf42e-40b0-4db8-9748-774751840a14",
"name": "Note adhésive",
"type": "n8n-nodes-base.stickyNote",
"position": [
-720,
-720
],
"parameters": {
"width": 660,
"height": 900,
"content": "# GitHub Push → README & CHANGELOG check\n\n## 1) Webhook (receive push from GitHub)\n* **Node:** Webhook (POST)\n* **GitHub setup:** Repository → *Settings → Webhooks → Add webhook*\n * **Payload URL:** `https://<your-n8n-domain>/webhook/github/push`\n * **Content type:** `application/json`\n * **Event:** *Just the push event*\n * **Secret:** optional\n * **Branches:** choose specific branches or send all\n## 2) IF (filter)\n* **Node:** IF\n* **Checks:** push contains `README` and `CHANGELOG` in *added* or *modified* files\n* **True:** continue\n* **False:** do nothing (stop)\n"
},
"typeVersion": 1
},
{
"id": "b08176e7-283f-4ed3-8c54-c90daeb05d59",
"name": "Note adhésive1",
"type": "n8n-nodes-base.stickyNote",
"position": [
48,
-720
],
"parameters": {
"color": 4,
"width": 640,
"height": 920,
"content": "# GitHub → Extract → Merge/Aggregate\n\n## 1) GitHub (Get Repository File)\n* **Credentials:** GitHub OAuth2 or PAT with read access.\n* **Owner/Repo/Branch:** from webhook payload (`body.repository`, branch from `body.ref`).\n* **File Path:** `README.md` and `CHANGELOG.md` (use two GitHub nodes).\n* **Output:** binary files.\n## 2) Extract from File (text)\n* **Operation:** `text`.\n* **Input:** binary from each GitHub node.\n* **Output:** plain text for README and CHANGELOG.\n## 3) Merge & Aggregate\n* **Merge:** *Wait for Both* (combine README + CHANGELOG into one item).\n* **Aggregate:** `aggregateAllItemData` (single item with `documents` list holding both contents).\n"
},
"typeVersion": 1
},
{
"id": "90e42f76-4ca3-4e6d-902c-2b4f1e29c0a4",
"name": "Note adhésive2",
"type": "n8n-nodes-base.stickyNote",
"position": [
752,
-720
],
"parameters": {
"width": 720,
"height": 920,
"content": "# LLM → Post to Twitter & LinkedIn\n\n## 1) LLM (OpenAI → JSON)\n* Nodes: OpenAI Chat Model → Basic LLM Chain (+ Structured Output Parser)\n* Input: aggregated README + CHANGELOG\n* Output: JSON with `\"twitter\"` and `\"linkedin\"`\n## 2) Publish\n* **Twitter node**\n * Auth: credentials\n* **LinkedIn node (Person)**\n * Auth: credentials\n * Required scope: `w_member_social`, `openid`\n"
},
"typeVersion": 1
},
{
"id": "3f31d48b-c0cc-4140-9bd8-02f4267f7f65",
"name": "Note adhésive3",
"type": "n8n-nodes-base.stickyNote",
"position": [
-1520,
-720
],
"parameters": {
"color": 5,
"width": 780,
"height": 900,
"content": "# Generate social posts from GitHub pushes to Twitter and LinkedIn\nOn each GitHub *push*, this workflow checks if the commit set includes **README.md** and **CHANGELOG.md**, fetches both files, lets an **LLM** generate a Twitter and LinkedIn post, then publishes to **Twitter** and **LinkedIn (Person)**.\n## Apps & Nodes\n* **Trigger:** Webhook\n* **Logic:** IF, Merge, Aggregate\n* **GitHub:** Get Repository File (×2)\n* **Files:** Extract from File (text) (×2)\n* **AI:** OpenAI Chat Model → LLM Chain (+ Structured Output Parser)\n* **Publish:** Twitter, LinkedIn (Person)\n## Prerequisites\n* **GitHub:** OAuth2 or PAT with repo read.\n* **OpenAI:** API key.\n* **Twitter:** OAuth2 app with *Read and Write*; scopes `tweet.read tweet.write users.read offline.access`.\n* **LinkedIn (Person):** OAuth2 credentials; **required scope:** `w_member_social`, `openid`.\n## Setup\n1. **GitHub Webhook:** Repo → *Settings → Webhooks*\n\n * Payload URL: `https://<your-n8n-domain>/webhook/github/push`\n * Content type: `application/json` • Event: *Push* • Secret (optional) • Branches as needed.\n2. **Credentials:** Connect GitHub, OpenAI, Twitter, and LinkedIn (Person).\n## How it Works\n1. **Webhook** receives GitHub push payload.\n2. **IF** checks that `README` and `CHANGELOG` appear in *added/modified*.\n3. **GitHub (Get Repository File)** pulls `README.md` and `CHANGELOG.md`.\n4. **Extract from File (text)** converts both binaries to text.\n5. **Merge & Aggregate** combines into one item with both contents.\n6. **LLM (OpenAI + Parser)** returns a JSON with `twitter` and `linkedin`.\n7. **Twitter** posts the tweet.\n8. **LinkedIn (Person)** posts the LinkedIn text.\n"
},
"typeVersion": 1
},
{
"id": "95134ca6-76d8-49a2-ae2c-6f1d7121c9a6",
"name": "Extraire du fichier README",
"type": "n8n-nodes-base.extractFromFile",
"position": [
224,
-208
],
"parameters": {
"options": {},
"operation": "text"
},
"typeVersion": 1
},
{
"id": "a61fd00a-34f2-4e8b-bdf5-7e9c686f58ed",
"name": "Extraire du fichier CHANGELOG",
"type": "n8n-nodes-base.extractFromFile",
"position": [
224,
32
],
"parameters": {
"options": {},
"operation": "text"
},
"typeVersion": 1
}
],
"active": true,
"pinData": {},
"settings": {
"executionOrder": "v1"
},
"versionId": "78f3c247-4dc6-474b-8a51-3a55a072dd92",
"connections": {
"9229a69b-fe35-4a25-a8d8-bc3e1f6d8894": {
"main": [
[
{
"node": "b3ae8e15-dd37-4207-a78c-4b18b72c9cb8",
"type": "main",
"index": 0
},
{
"node": "a19b66f5-0927-4c59-8343-f2cea29feb5e",
"type": "main",
"index": 0
}
],
[
{
"node": "4e9ebbbd-9c37-4dbe-a9b6-dc0f72a85877",
"type": "main",
"index": 0
}
]
]
},
"a991ef0a-48ef-4ccf-98c6-587edddafa96": {
"main": [
[
{
"node": "e35435d4-f3f8-44f1-ac78-21efda7076ac",
"type": "main",
"index": 0
}
]
]
},
"29f468dd-8822-4de7-9bf7-0205e8b51b12": {
"main": [
[
{
"node": "9229a69b-fe35-4a25-a8d8-bc3e1f6d8894",
"type": "main",
"index": 0
}
]
]
},
"e35435d4-f3f8-44f1-ac78-21efda7076ac": {
"main": [
[
{
"node": "82c95a95-4a00-405d-a826-9c387a3f76be",
"type": "main",
"index": 0
}
]
]
},
"a19b66f5-0927-4c59-8343-f2cea29feb5e": {
"main": [
[
{
"node": "95134ca6-76d8-49a2-ae2c-6f1d7121c9a6",
"type": "main",
"index": 0
}
]
]
},
"b3ae8e15-dd37-4207-a78c-4b18b72c9cb8": {
"main": [
[
{
"node": "a61fd00a-34f2-4e8b-bdf5-7e9c686f58ed",
"type": "main",
"index": 0
}
]
]
},
"82c95a95-4a00-405d-a826-9c387a3f76be": {
"main": [
[
{
"node": "c2454817-d07d-45e0-8c5f-03a68c619eb8",
"type": "main",
"index": 0
},
{
"node": "944f1559-f79e-4034-b70f-0246c8b40a5f",
"type": "main",
"index": 0
}
]
]
},
"927afa09-65bf-4c62-955c-42a3bb098f28": {
"ai_languageModel": [
[
{
"node": "82c95a95-4a00-405d-a826-9c387a3f76be",
"type": "ai_languageModel",
"index": 0
}
]
]
},
"95134ca6-76d8-49a2-ae2c-6f1d7121c9a6": {
"main": [
[
{
"node": "a991ef0a-48ef-4ccf-98c6-587edddafa96",
"type": "main",
"index": 0
}
]
]
},
"43f2f142-8319-445f-80a9-61c56ae6ddad": {
"ai_outputParser": [
[
{
"node": "82c95a95-4a00-405d-a826-9c387a3f76be",
"type": "ai_outputParser",
"index": 0
}
]
]
},
"a61fd00a-34f2-4e8b-bdf5-7e9c686f58ed": {
"main": [
[
{
"node": "a991ef0a-48ef-4ccf-98c6-587edddafa96",
"type": "main",
"index": 1
}
]
]
}
}
}Comment utiliser ce workflow ?
Copiez le code de configuration JSON ci-dessus, créez un nouveau workflow dans votre instance n8n et sélectionnez "Importer depuis le JSON", collez la configuration et modifiez les paramètres d'authentification selon vos besoins.
Dans quelles scénarios ce workflow est-il adapté ?
Avancé - IA Multimodale
Est-ce payant ?
Ce workflow est entièrement gratuit et peut être utilisé directement. Veuillez noter que les services tiers utilisés dans le workflow (comme l'API OpenAI) peuvent nécessiter un paiement de votre part.
Workflows recommandés
Jorge Martínez
@jorgemartinezjamAI Engineer & Full Stack Developer skilled in designing intelligent agents, workflow automation, and NLP solutions. Experienced with LLMs, Relevance AI, make, zapier, and n8n. Proficient in integrating AI and chatbots into web apps, optimizing processes, and leveraging automation for real-time productivity. Strong background in backend and frontend development, REST APIs, and database management.
Partager ce workflow