ai-video-marker workflow
A workflow for short AI videos: nine flows, agent roles, skills and Python tools for pixel animation.
- 526 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- The agent runs the repository Python scripts and writes videos and service files into `outputs/`.
- Dependencies install through `pip` from `requirements.txt`.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add workflows/ai-video-markerDetects 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 the repository Python scripts and writes videos and service files into `outputs/`.
- Dependencies install through `pip` from `requirements.txt`.
This is third-party code. Review the repository files before installing.
What it does
ai-video-marker describes how to take a clip from idea to export with an AI assistant: idea, storyboard, assets, animation, music and export. The repository holds nine content flows (six main and three candidates), two helper chains, nine agent roles such as directing and planning or sound and post-production, ten skills, three runnable pixel scenes, a finished comic and 18 text starters. To work with it, open the whole repository in an AI assistant that reads files and runs commands. A newcomer makes a 36-second pixel short about a robot and a lighthouse: `python tools/make_video.py` builds the video with no API keys. The tools include `doctor.py` for environment checks, `check_video.py`, `contact_sheet.py` and `asset_library.py` for checking and finding assets. Methods for Blender layered 2.5D animation and for editing in CapCut (Jianying) are described as well. The documentation is written in Chinese.
Who it is for. Authors of short clips and developers who want to assemble video with Claude Code or Codex through a ready process.
Good fit when
- A short pixel clip is needed and a ready scene with a local render will do.
- A video idea must become a storyboard and a plan by role: director, artist, sound.
- A comic-based clip must reach editing in Jianying with a fixed process.
- Work on a video must continue in a new agent window with the same context.
Not a fit when
- One-command installation into the agent skills directory is wanted: there is no ready command.
- Reading Chinese documentation is not an option.
- The project must run on Windows or Linux: only macOS has been tested.
Example request
Use the repository assets and make a short pixel clip that differs from the README example: preview first, then a full export.Limitations
This is a set of materials and tools with no package: the repository is cloned and opened whole in an agent. Python 3.9 or newer is required, dependencies install from `requirements.txt`, Pillow draws frames and `imageio-ffmpeg` provides FFmpeg. Image, voice and image-to-video generation and Jianying editing stay with your assistant and local environment. The authors write that a run was verified on macOS and Windows and Linux are not verified yet. Documentation and role names are in Chinese. The last commit in the repository is from November 2026.
How to disable. Delete the cloned repository folder and the `.venv` virtual environment inside it.
FAQ
How do I start with ai-video-marker?
Clone the repository, open it whole in an agent with file and command access and ask for a video from the repository assets. By hand: `python3 -m venv .venv`, `pip install -r requirements.txt`, `python tools/make_video.py`.
Are API keys needed?
Not for the starter clip: Pillow draws frames and FFmpeg comes from `imageio-ffmpeg`. Image and voice generation go through your assistant.
What is in the repository?
Nine video flows, two helper chains, nine agent roles, ten skills, three pixel scenes, a comic and 18 text starters.
How do I check a finished video?
With `python tools/check_video.py outputs/my-first-video/video.mp4 --manifest outputs/my-first-video/run.json`, and `contact_sheet.py` builds a contact sheet of frames.
Does it work on Windows?
The authors verified a run on macOS. For Windows they give environment commands with `py -3 -m venv .venv`, but operation there was not verified.
Discussion
No questions or reviews yet.
Related
WorkflowsA skill for Cursor, Claude Code and Codex: from a product name or URL it builds a 4K promo with music, voice-over and checks
Claude Code directs a video: transcription, script, HyperFrames scenes, subtitles and Remotion assembly with a localhost dashboard
Five skills for explainer videos: script, storyboard, narration with captions, animated diagrams and whiteboard drawing.
A set of skills and local apps for recording, rough-cutting and animating a YouTube video with an AI agent.
Open Film Skills bundles 22 skills for AI film: screenwriting, direction, assets, storyboards, prompts and post-production.
rea-video is an agent skill that breaks a video down shot by shot, calling out real versus AI footage and building a remake plan.
README in short
A ready AI video workflow: roles, skills and verifiable tools in one repository. The starter clip needs no API keys.