cinecode skill
The agent writes the script, renders every frame in code with headless Chrome and returns an MP4: shorts, ads, explainers.
- 2,387 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- The agent runs ffmpeg and headless Chrome and writes project files.
- Narrator keys are sent to external voice services.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/cinecodeDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The agent runs ffmpeg and headless Chrome and writes project files.
- Narrator keys are sent to external voice services.
This is third-party code. Review the repository files before installing.
What it does
The `cinecode` skill leads an agent from a one-line idea to a finished video. The agent proposes a plot by crossing the idea with a real object from a museum, nature or history, and checks it against films already made. The user picks a look from a board of stills and approves the first two scenes. Frames are rendered in code through headless Chrome (three.js, 2D, paper, pixels and other techniques), sound is assembled separately and ffmpeg joins the final cut. Modes cover short films, long YouTube videos, ads and vertical 30–60 second explainers.
Who it is for. Video creators and marketers who need a clip without shooting or editing software.
Good fit when
- An ad about a product is needed with your colours, fonts and logo.
- A short film idea exists but there is no studio or crew.
- A YouTube explainer is in the works where every claim about people and events cites a source.
- Social networks need a half-minute vertical video with diagrams and large type.
Not a fit when
- Footage of real people or live shots is required.
- The video is due today and rendering from code takes a lot of time.
- The agent only works in a browser chat without file access.
Example request
Make a 40-second ad for our service, build the frames in code, offer three looks on a board of stills and wait for my pick.Limitations
The skill is built for Claude Code and Cursor on your own computer. It needs ffmpeg and headless Chrome; on Windows without git the agent offers to install it through `winget` after your consent. A narrator is optional and runs on your own Gemini or ElevenLabs key or a free Turkish voice. The main documentation is in Turkish, with English, Chinese and Japanese translations. Long films are built in acts and chapters, which stretches the work over many steps and token spend.
How to disable. Delete `~/.claude/skills/cinecode` (`%USERPROFILE%\.claude\skills\cinecode` on Windows). Revoke the Gemini and ElevenLabs keys in their dashboards.
FAQ
What kinds of video can it make?
Short films, YouTube videos of ten minutes and more, ads and vertical social explainers of 30–60 seconds.
Are paid services required?
No. A paid service is not required; a narrator is optional through your own Gemini key, ElevenLabs or a free Turkish voice.
How do I update the skill?
Tell the agent "update cinecode" and it pulls the newest version. Do not install it in two places, or the agent will see two copies.
Which phrases trigger it?
It starts on everyday requests like "make a short film", "product video" or "ad", plus Turkish phrasings.
Which styles are available?
3D with real light, 2D animation, paint, ink, paper, clay, motion graphics and pixels. The look is chosen from a board of stills from real films.
Discussion
No questions or reviews yet.
Related
SkillsA skill that makes promo videos from code: direction, in-browser animation with motion blur, its own score and a 60 fps MP4
A skill and CLI that let Claude Code or Codex make videos as web pages, with the Studio editor to change the result
A skill that makes your agent film a short product video: concept, edit, sound and render on Remotion or HyperFrames
A skill that makes video in pure code: narrated explainers and website promos in 35 visual styles, in 16:9 and 9:16.
The agent scripts, voices and draws a narrated documentary in code, with one continuous camera over photos, maps and cards.
A Claude Code skill for code-made video: it splits a clip into scenes and picks an engine for each, from Paper.js to Manim.
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| cinecode (this card) | no | 2,387 | |
| claude-video-kit | no | 124 | builds a vertical 9:16 clip from a script on Remotion; cinecode draws frames in code and goes from idea to long films |
| Claude Remotion Skill | no | 984 | teaches an agent to make video on Remotion; cinecode sets a full process from plot to sound |
README in short
Every frame is drawn in code and rendered in headless Chrome, while separate steps and ffmpeg handle sound and the final cut. The look is chosen on a board of stills and the user approves the first scenes.