The Viral Cloning Protocol: Copy Paste TikTok Trends with n8n & Gemini 3

Stop trying to reinvent the wheel. The most efficient way to scale short-form content isn’t “originality” but replication. Agencies are currently charging five figures to run trend-jacking operations that you can build in a single afternoon.

We are deploying a sophisticated automation architecture using n8n. The objective: Take a URL from a viral TikTok, strip the watermark, use Gemini 3’s multimodal capabilities to reverse-engineer the prompt, and regenerate a compliant clone using Wan 2.5. Finally, we auto-deploy to every social grid simultaneously.

This is the exact infrastructure used to ride algorithmic waves without ever turning on a camera.

The Stack

• n8n: The orchestration layer.
• Gemini 3: The vision model (analysis).
• Wan 2.5 (via Kie AI): The video generation engine (less censorship than Sora).
• Blotato: The distribution node.

---

Step 1: The Input Vector (n8n Form)

We need a trigger. A simple webhook isn’t enough; we want a user interface to paste the target URL.

1. Create a new n8n workflow.
2. Add the n8n Form Trigger node.
3. Configuration:
   • Form Title: Viral Clone Request
   • Form Fields: Add a Text Field labeled URL. Set it to Required.
4. Execute Workflow: Open the test URL, paste a TikTok link (e.g., a viral dance or skit), and hit submit.

This passes the raw TikTok URL into the workflow’s memory.

Step 2: The Scrape (Bypassing the Gatekeepers)

TikTok aggressively blocks scrapers. To get the raw, watermark-free video, we have to spoof a real user browser.

1. Add an HTTP Request node.
2. Method: GET.
3. URL: Map the URL from the Form Trigger.
4. Headers (CRITICAL): You must manually add a User-Agent header to trick TikTok into thinking you are a standard browser. Use this string:
   Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36
5. The Extraction Code: You will receive a massive block of HTML. Add a Code Node to regex match the video_id and extracting the cookies.

Note: The extraction logic parses the HTML for the JSON blob containing the video download URL. This changes frequently, so ensure your regex targets the PlayAddr or video_url within the script tags.

Step 3: The Download

Once we have the cleaned URL from the scrape step:

1. Add another HTTP Request node.
2. URL: Map the cleaned video URL from the Code Node.
3. Headers:
   • Referer: https://www.tiktok.com/
   • Cookie: Map the cookies extracted in the previous step.
4. Response Format: Select File (Binary).

You now have the viral video sitting in your n8n RAM, stripped of watermarks.

Step 4: The Brain (Gemini 3 Analysis)

This is where Gemini 3 shines. Unlike older models, Gemini 3 has native video understanding. We don’t want a summary; we want a production prompt.

1. Add the Google Gemini node.
2. Model: gemini-1.5-pro-preview (or Gemini 3 if available in your version).
3. Input: Binary (connect the video from Step 3).
4. The Prompt: This needs to be surgical. Paste this exact prompt:

“Analyze this video moment by moment. Describe the video to an AI video generator so I can replicate this video exactly including the audio. Output ONLY a prompt for an AI video generator. Prompt can be no more than 800 characters.”

Gemini will watch the video and output a text prompt describing the lighting, camera angle, action, and subject (e.g., “Vertical smartphone footage of a Michael Jackson impersonator eating fried chicken…”).

Step 5: The Engine (Wan 2.5 Generation)

We are using Wan 2.5 because it has significantly fewer “safety” refusals than OpenAI’s Sora. If you try to clone a celebrity or a copyrighted character on Sora, it blocks you. Wan usually executes.

We will access Wan 2.5 via Kie AI (an API wrapper).

1. Add an HTTP Request node.
2. Method: POST.
3. URL: https://api.kie.ai/v1/jobs/createTask
4. Authentication: Header Auth (Authorization: Bearer YOUR_KIE_API_KEY).
5. JSON Body:

{
  "model": "wan/2-5-text-to-video",
  "input": {
    "prompt": "{{ $json.output_from_gemini }}",
    "aspect_ratio": "9:16",
    "duration": "10",
    "resolution": "720p",
    "enable_prompt_expansion": false
  }
}

Note: We disable prompt expansion because Gemini 3 has already optimized the prompt.

Step 6: The Async Loop

Video generation is not instantaneous. We need a logic loop to wait for completion.

1. Wait Node: Set to 3 minutes.
2. HTTP Request (Check Status):
   • Method: GET.
   • URL: https://api.kie.ai/v1/jobs/recordInfo?taskId={{ $json.taskId }}
3. If Node (Logic):
   • Condition: state EQUALS success.
   • True: Proceed to upload.
   • False: Loop back to the Wait Node.

Step 7: The Distribution (Blotato)

Blotato is a purpose-built tool for social automation. It handles the API nightmares of TikTok, Instagram, and YouTube.

1. Install Community Node: Go to Settings > Community Nodes > Install n8n-nodes-blotato.
2. Upload Node:
   • Action: Upload Media.
   • Input: The video URL returned by Kie AI in the success step.
   • This returns a Blotato-hosted public URL.
3. Post Nodes:
   • Add separate nodes for Instagram, TikTok, and YouTube Shorts.
   • Action: Create Post.
   • Media URL: Map the Blotato URL.
   • Caption: Use a quick AI node to generate hashtags based on the original prompt, or hardcode generic viral tags like #fyp.

The Result

You paste a URL. Five minutes later, a legally distinct, AI-generated clone of that video appears on your Instagram Reels, TikTok, and Shorts. You have successfully detached your time from your content output.