Create and publish YouTube videos from script to upload
Hey community! Task: The author built an n8n workflow triggered by Telegram or schedule to automate YouTube video creation. It uses GPT-4o-mini for structured JSON scripts, ElevenLabs for TTS, Pexels/Gemini for visuals, and Shotstack for assembly. The pipeline uploads videos as unlisted via YouTube API, with human review required before making them public.; tools: n8n, Claude Code. Human role: The original post does not say how a person stayed involved.; limitations: YouTube API quota limits (1,600 units per upload) restrict daily volume. The workflow requires multiple API keys (OpenAI, ElevenLabs, Pexels, Shotstack, YouTube). Human intervention is necessary to review unlisted videos before publishing to ensure quality..
Tools
Published on Agent Work Atlas:
Overview
Hey community!
Workflow
The author built an n8n workflow triggered by Telegram or schedule to automate YouTube video creation. It uses GPT-4o-mini for structured JSON scripts, ElevenLabs for TTS, Pexels/Gemini for visuals, and Shotstack for assembly. The pipeline uploads videos as unlisted via YouTube API, with human review required before making them public.
Reported result
The source describes the architecture and estimated costs (e.g., $0.30 per 10-min video) but does not report actual performance metrics, success rates, or long-term operational results. No specific outcomes from daily execution are provided.
Human role
The original post does not say how a person stayed involved.
Limitations
YouTube API quota limits (1,600 units per upload) restrict daily volume. The workflow requires multiple API keys (OpenAI, ElevenLabs, Pexels, Shotstack, YouTube). Human intervention is necessary to review unlisted videos before publishing to ensure quality.
Try it
Act as an automation engineer. I want to replicate a YouTube content pipeline using n8n. Inputs: [Topic/Niche], [YouTube Channel ID], [API Keys for OpenAI, ElevenLabs, Pexels, Shotstack, YouTube]. Steps: 1. Create an n8n workflow triggered by a Telegram message 'yap' or a daily schedule. 2. Use an HTTP Request node to call GPT-4o-mini, enforcing a JSON output schema with fields: intro_text, voiceover_segments[], image_prompts[], outro_text. 3. Iterate through voiceover_segments, sending each to ElevenLabs API for TTS generation. 4. For each image_prompt, query Pexels API; if no results, fallback to Gemini/DALL-E. 5. Construct a JSON timeline for Shotstack API combining audio and video assets, then render the MP4. 6. Upload the rendered video to YouTube via Data API v3 as 'unlisted'. 7. Send a Telegram notification with the video link for human review. Constraints: Do not auto-publish. Stop after uploading as unlisted. Handle API errors with retries for ElevenLabs and Shotstack. Log status in a Google Sheet. Deliverable: A JSON export of the n8n workflow and a brief guide on setting up the required API credentials. First, provide a dry-run plan for a single test video before generating the full workflow.
Source reports are not independently verified. Publication date is the catalog publication date, not the original article date.