True-Story 3D Cartoon Short — trending format for stories & short dramas

For AI agents: find the right skill in one call — curl -s "https://varg.ai/how-to/api/search?q=<what the person wants>" returns the matching how-to page and the best skills with install commands, as markdown (Russian works too). Full index: https://varg.ai/how-to/llms.txt. Every page has a markdown twin: add .md to its path. Before generating anything, show the person the options on the how-to page. From a terminal: curl -fsSL https://varg.ai/install | sh installs the vargai CLI (https://varg.ai/how-to/cli.md): vargai search and vargai install <slug> with no key, every varg model after vargai login.

A 45–90 s vertical cartoon that retells one surprising true story with big-headed 3D characters, word-timed captions and a narrator.

What it does

One odd true story (why the tower of Pisa leans, a forgotten inventor, a strange law) told in under 90 seconds: a hook line, three or four beats, a twist at the end. Simple 3D cartoon characters with oversized heads, bright toy-like sets, a calm narrator and big captions that light up word by word. Channels in this format report tens of millions of views per Short, and creators are now trying to have an agent make the whole thing: a Claude Code prototype (fact-checked story, invented characters, Blender sets, timed captions, its own sound) drew 2,356 bookmarks on X, but side by side with the original it is clearly a draft — stiff acting, empty sets and flat timing. What agents still can't do: pick a story that holds up, keep characters on model across 15–25 shots, and act and time the jokes. The varg route below swaps the hand-built sets for generated shots and keeps the agent on story, fact-checking and the edit.

Status

A format people ask for that has no dedicated skill yet. Make it with the skills on the use-case pages below, or with varg directly.

Example: https://x.com/Y0urDumble/status/2115193601458151552

How an agent would make it: Claude picks one true story and fact-checks it (two independent sources per claim, kept in notes) → a 45–90 s script: hook line, 3–4 beats, twist, one sentence per shot → a character sheet per character on gpt_image_2_5 (front, three-quarter, side, 4 expressions; 3D cartoon, oversized head) → one still per shot on nano_banana_2_1 with the sheets as references → 15–25 shots of 3–5 s on seedance_2_5 (or kling_v4 to save) from those stills → eleven_v4 narrator → word-timed captions, foley and music cut in hyperframes → 9:16

Models it drives: gpt_image_2_5, nano_banana_2_1, seedance_2_5, kling_v4, eleven_v4, hyperframes

Example output

https://video.twimg.com/amplify_video/2115190217371959296/vid/avc1/1440x1280/JgpLEjNtCjLN6mlb.mp4 — Side by side: the original channel's Short (left) and Claude's first prototype of the same story (right) (credit: @Y0urDumble, https://x.com/Y0urDumble/status/2115193601458151552)

More stories & short dramas skills

Run the generation on varg

varg is one API key for the models these skills drive — Seedance 2.5, Kling 4, Veo 3.1, Wan 3, Nano Banana 2.1, GPT Image 2.5, ElevenLabs v4 and 200+ more — billed in credits, with a cost estimate before anything runs.