AI NewsOpen sourceAnnouncement
LearnPrompt's awesome-seedance passed 1,332 stars in 29 days, and installs as a Claude Code or Codex skill for ByteDance's Seedance video model
LearnPrompt's awesome-seedance repository passed 1,332 stars in the 29 days since it was created, offering 463 verified prompt cases, 25 copy-ready templates and 60 installable skills for the ByteDance Seedance 2.5 and 2.0 video models, all under MIT.
Image: GitHub
Why it mattersA creator making short videos with Seedance now has 463 prompts other people have already tested and a set of Claude Code or Codex skills that turn a short brief into the full prompt, so the first draft starts from a known-working pattern.
A creator working with ByteDance's Seedance video model can now install a Claude Code or Codex skill that already knows the vendor's prompt shape, the mistakes other people ran into, and 463 case studies traced back to their original posts. The repository is LearnPrompt's awesome-seedance, and it passed 1,332 stars on the 29th day after it was created.
What the repository contains
LearnPrompt's own count is 463 verified cases, 264 cross-model retest runs with public verdicts, 25 copy-ready templates and 60 installable AI-video skills, in English, Chinese and Japanese. The templates group by task: structural foundations, realism, commercial work, narrative, animation and action. Each template lists the pattern that worked and the mistake other people made on the same task.
The verified cases are traced back to the goodcase.ai library, so a reader can open the original post and see the finished clip before installing anything.
How the skills install
The 60 skills are packaged as instruction files for a coding agent. LearnPrompt lists Claude Code and Codex as the supported harnesses and provides an install script that copies the skill files into the agent's own skill directory. A user then invokes a skill by name inside the agent and passes a short brief; the skill fills in the vendor-specific structure that the case library has already tested.
The repository is under MIT for the code and CC BY 4.0 for the curated material.
What Seedance is
Seedance is ByteDance's text-to-video model, sold in China as Jimeng and internationally as Dreamina. The 2.5 and 2.0 releases are the two versions this library targets. Both are commercial models with their own paid tiers; the repository is only the prompt work, and using it still requires a Seedance account.
Why the star count matters
The repository was created on 26 August. A daily rate of about 46 stars over the 29 days since is a real signal of adoption on GitHub, above the level of a small side project. Combined with the multilingual documentation and the fact that the whole thing is packaged as an installable skill for a coding agent, it suggests video-prompt work has become something a creator adds to their agent once and re-uses, with a shared library of tested patterns to pull from.
One caveat is worth stating plainly. This is one team's curation of another company's model. If ByteDance changes the prompt format in a 3.0 release, the 25 templates and the 264 retest runs will need to be redone. That is the current cost of a curated library: it is tied to the vendor's current interface, and it ages when the vendor changes it.
Source
- Repository: LearnPrompt/awesome-seedance
- Verified case library: goodcase.ai
This item was written by an AI system from the linked source. Reveneau is responsible for what it publishes.
Get AI News in your inbox
New developer tools, model and agent releases, and how teams are actually using them to release software. Short, and only when there is something worth reading.