cutcli-cookbook
Open-source cookbook, JSON templates, AI prompts, and docs for cutcli — the CapCut / Jianying (剪映) draft CLI. Generate editable video drafts from code, Cursor, Claude Code or any MCP agent.
cutcli generates standard CapCut / Jianying drafts that the desktop app can open directly. This repo collects ready-to-run examples, reusable JSON templates, AI prompts, and the official documentation source. Contributions welcome.
30-second start
# 1. Install cutcli
curl -s https://cutcli.com/cli | bash
# 2. Run the simplest example
git clone https://github.com/xuliang2024/cutcli-cookbook.git
cd cutcli-cookbook/examples/01-hello-caption
bash run.sh
# 3. Open CapCut / Jianying — the new draft is already in the list
Full guide: https://docs.cutcli.com/guide/installation
New in cutcli v1.3.9: cloud rendering
cutcli can now upload a local draft zip and submit it to the cutcli.com cloud renderer. Use it for smoke tests, preview rendering, or scheduled draft generation without opening CapCut locally.
# Save your cloud API key once
cutcli auth set --api-key cut_live_xxx_yyy
# Render an existing local draft in the cloud
cutcli cloud render <draftId> --pretty
# Inspect queue/results
cutcli cloud jobs --status queued --pretty
cutcli cloud job <renderJobId> --pretty
# Run one scheduled render cycle for automation checks
cutcli timer render --count 1 --pretty
Reference: https://docs.cutcli.com/reference/cli
What's inside
| Directory | Purpose |
|---|---|
examples/ | One-shot runnable examples (each with run.sh + README.md + preview.gif) |
templates/ | Reusable JSON snippets (caption / animation / filter presets) |
prompts/ | Prompts for Cursor / Claude / ChatGPT to drive cutcli |
docs/ | VitePress source for docs.cutcli.com |
showcase/ | Monthly community showcase |
worker/ | Cloudflare Worker (R2 reverse proxy with SPA fallback) |
Featured P0 examples
| Example | Demo | Length |
|---|---|---|
| 01-hello-caption | One caption + entrance animation | 5 s |
| 02-image-slideshow-bgm | 3-image slideshow + transitions + BGM | 9 s |
| 03-tiktok-keyword-highlight | Multiple captions with keyword highlight | 6 s |
| 04-easy-by-audio | cutcli draft easy auto-fits to audio | adaptive |
| 05-keyframe-zoom-in | Image + keyframe zoom | 5 s |
FAQ
What is cutcli?
cutcli is a single-binary command-line tool that turns shell commands or JSON into a standard CapCut / Jianying (剪映) draft folder. Open CapCut on desktop and the draft is already in your draft list — fully editable, with captions, animations, transitions, audio and keyframes intact. No reverse-engineered file format, no manual JSON patching.
Install: curl -s https://cutcli.com/cli | bash
Can cutcli render drafts in the cloud?
Yes. Since cutcli v1.3.9, cutcli cloud render <draftId> can zip, upload, and submit a local draft to the cloud renderer. cutcli cloud jobs, cutcli cloud job, and cutcli cloud result show queue status and final video URLs. cutcli timer render can run the same flow on an interval for automation checks.
What's in this cookbook?
examples/— copy-paste runnable cases (one caption, image slideshow, TikTok keyword highlight, audio-driven layout, keyframe zoom)templates/— reusable JSON snippets for captions / animations / filtersprompts/— system prompts for Cursor, Claude Code, ChatGPT and MCP agents to drive cutclidocs/— VitePress source for docs.cutcli.comworker/— Cloudflare Worker (R2 reverse proxy + SPA fallback) that powers the docs site
How do I use cutcli with Cursor, Claude Code or an MCP agent?
cutcli is just a shell command, so any AI coding agent can drive it. We ship ready-made system prompts in prompts/ — drop them into your agent and ask things like "add a 3-second caption that fades in at 1s" or "build a 9-second slideshow from these three images with BGM". The agent emits cutcli ... calls and you get a CapCut draft. See the integration guide at https://docs.cutcli.com/guide/ai-integration.
Does cutcli work with both CapCut (international) and Jianying (剪映 Chinese)?
Yes. The draft format is the same on both apps; cutcli writes drafts that either client can open directly from its draft folder.
How is this different from manually editing CapCut JSON?
Hand-writing CapCut draft JSON means dealing with microsecond timestamps, undocumented enums, normalized 0–1 coordinates and a half-dozen sibling files that must stay in sync. cutcli handles all of that and exposes a stable CLI surface (cutcli draft create, cutcli captions add, cutcli draft easy, …) that survives CapCut version bumps. Same reason people use FFmpeg instead of writing MP4 boxes by hand.
Contributing
We welcome every kind of contribution:
- Add a new example (easiest — copy a folder and edit, no setup required)
- Translate or improve docs
- Fix a bug
- Submit your work to
showcase/
See CONTRIBUTING.md.
Links
- Website: https://cutcli.com
- Docs: https://docs.cutcli.com
- Install:
curl -s https://cutcli.com/cli | bash - Discussion: Issues / Discussions
License
MIT © 2026 m007 and cutcli-cookbook contributors