
capcut-cli
CLI在终端里创建和修改剪映 / CapCut 本地草稿,可按静音粗剪、加带样式字幕、生成多语种版本,结果仍能在剪映里逐轨精修,适合课程和自媒体视频。需要 Node 18+ 与桌面版剪映或 CapCut,字幕需 Whisper,翻译需 Anthropic API Key;剪映 6.0 起的加密草稿读不了。非字节官方项目。
capcut-cli
English | 中文
Create and edit real CapCut / JianYing projects from the terminal — or any LLM agent.
Open the result in CapCut with every track still editable. capcut-cli works directly on the local draft store: JSON in, JSON out, with no upload, API, MCP server, or HTTP daemon.
Also from the maintainer: 500 Years Frozen in Time: The Inca Children Mystery, a 14:49 documentary from Bronze Age Banter.
raw recording → silence-aware cuts + styled captions → editable CapCut / JianYing draft
▶ Captioned output example (60 seconds)
https://github.com/user-attachments/assets/4e6ee99c-0745-4cfb-8e9b-ad873fb1259b
Install and open your first editable draft
Prerequisites: Node ≥ 18 (built-ins only — no native modules). Optional tools unlock specific commands: Whisper for caption, FFmpeg for render, ffprobe for automatic media metadata, and ANTHROPIC_API_KEY for translate.
npm install -g capcut-cli
From Python: pip install capcut wraps the same binary — capcut.run("quickstart", "my-short", video="clip.mp4", ratio="9:16") — see python/README.md.
capcut doctor
capcut quickstart my-first --video clip.mp4 --srt captions.srt
capcut info ./my-first/ -H
Result: a real local project with video and captions on editable tracks — not a flattened export. Open it in CapCut or JianYing to review, adjust, and render. The publish click stays human.
Useful? Star capcut-cli to help other editors and agent builders find it.
For more practical tools for AI agents, from video automation to checks before they ship, follow René on GitHub.
Build from source instead: git clone https://github.com/renezander030/capcut-cli && cd capcut-cli && npm install && npm run build (then npm link to expose capcut). Or run any command without installing: npx capcut-cli <command>.
[!IMPORTANT] Upgrade before using older versions. Fixture bundles made with versions up to 0.17.2 may contain stable device identifiers and must be treated as unsanitised (#59). Versions up to 0.17.0 also contain local command/filter injection paths and unsafe temporary-file or credential-output behaviour. These issues are fixed in 0.18.0 and 0.17.1 respectively. Run
npm install -g capcut-cli@latestand see the changelog for full details.
Disclaimer: This is an independent, community-maintained project. It is not affiliated with, sponsored by, or endorsed by CapCut, JianYing, or ByteDance Ltd. "CapCut" and "JianYing" (剪映) are trademarks of ByteDance Ltd. All product names, logos, and brands are the property of their respective owners and are used here only for identification (nominative) purposes.
An independent CLI for CapCut / JianYing that any LLM agent can drive — zero dependencies, no server, both namespaces in one binary.
JSON in, JSON out: every command reads and writes the local draft store directly, with no MCP server or HTTP daemon. On newer CapCut versions it detects and synchronizes every readable timeline target instead of assuming draft_content.json is the only source of truth. That gives any model (Claude, DeepSeek, GLM, Kimi) a deterministic boundary for inspection, building, subtitles, captions, translation, and long-form cuts.
Use it four ways:
- CLI —
npm install -g capcut-cli, thencapcut <command> <project> - Library —
import { loadDraft, lintDraft, saveDraft } from "capcut-cli"(typed, zero-dep) - Queue runner —
capcut servereads JSONL jobs from stdin, for n8n / Make / Coze - Agent sandbox (experimental) — build
capcut-core.wasmfor three read-only MCP tools with zero filesystem, network, environment, clock, random, stdio, or process imports
Give your agent the skill
One command installs the capcut-edit skill into Claude Code, Codex, Cursor, OpenCode and the other agents the skills installer supports:
npx skills add renezander030/capcut-cli
Claude Code can also load it as a plugin:
/plugin marketplace add renezander030/capcut-cli
/plugin install capcut-cli@capcut-cli
The skill teaches the agent every command, the progressive-disclosure habit (inspect first, never dump a whole draft), where the draft store lives on macOS and Windows, and the deterministic scripts for fades, Ken Burns and long-to-short cuts. It triggers on English and Chinese requests alike (剪映, 字幕, 草稿).
Capability-free Wasm tools for agents
Using an AI assistant with capcut-cli? Give it a safer “look, don’t touch” mode.
The optional Wasm tool lets an AI:
- describe what is inside a CapCut project;
- show what changed between two versions; and
- spot common timeline and caption problems.
It can examine only the project information you hand to it. It cannot browse your files, go online, read passwords or API keys, open other apps, or change the project.
Use it when you want an AI to review a draft with the least possible access. Skip it when you are using capcut-cli yourself or want the AI to edit or render—the normal CLI still does those jobs. This feature is experimental, optional, and changes nothing unless you set it up. Technical details and setup →
The experimental wasm/capcut-core source package moves the deterministic, JSON-in/JSON-out boundary into a WebAssembly Component:
inspectmatchescapcut infofor valid drafts.diffmatchescapcut difffor structural changes.lint-portableruns an explicit, parity-tested subset ofcapcut lintthat needs no host files or media probing.
The host reads a draft and passes its JSON as tool input. The component itself has no ambient capabilities, and CI proves the built world has zero imports before exercising all three functions through Wassette over MCP. From a source checkout, build it with npm --prefix wasm/capcut-core ci && npm run wasm:verify; setup and security details are in the component README.
Release notes
New in v0.30.0: per-command help (
capcut <command> --help,capcut help <command>); undo that survives an app save, with every restore checked against the write journal (restore,restore --listintegrity); and real word timings from whisper.cpp for karaoke and word-reveal captions, also readable throughcaption --words. Full details in the changelog.
New in v0.29.0: captions from any aligner's word timings without Whisper (
caption --words); reviewed batch plans bound to the draft (batch --plan/--apply-plan); a render fidelity census and output-duration check (render --strict/--verify); TTS pronunciation lexicons (tts --lexicon);segment-overlapandsegment-offscreenlint checks; andrestore --listnaming the command behind each step. Full details in the changelog.
For an existing nested project, use capcut diagnose <project> --active-timeline to inspect the selected document before editing. The opt-in follows the pointer on unverified builds and refuses invalid or conflicting selected documents; it does not bypass write guards.
To populate a project that CapCut already owns, create an empty project in the app and quit CapCut, then run:
capcut compile spec.json --into /path/to/app-created-project --active-timeline --check
capcut compile spec.json --into /path/to/app-created-project --active-timeline
Omit --active-timeline for a flat project without Timelines/. Every root and active mirror must be empty. The project keeps its name and registration; canvas and frame rate change only when supplied in the spec. Imported media is staged under the project, with registration and timeline backups committed together. Reopen, save, close and reopen in CapCut to check that edits persist. Automated safety tests cover these paths; desktop acceptance on your app build remains unverified.
Built with capcut-cli
- OpenChatCut — exports an agent-edited timeline, local media, audio, and captions into a real CapCut / JianYing draft for review and rendering.
Built something with capcut-cli? Open a showcase issue with a public link, one sentence about what it does, and an optional screenshot or demo.
Project descriptions are approved by their maintainers. Inclusion does not imply endorsement or affiliation.
Commands
JSON by default (pipe to jq); add -H for a human-readable table. Pass --jianying to use the JianYing enum namespace. Run capcut <command> --help for full flags.
| Group | Commands |
|---|---|
| Inspect | info · tracks · materials · version · lint |
| Browse / drill in | segments · texts · segment · material |
| Create | init · quickstart · compile (build a draft from a JSON spec) |
| Preview | render (low-res ffmpeg proxy — not CapCut's final render) |
| Add | add-video · add-audio · add-text (Wikimedia URLs supported, license-checked) |
| Edit / animate | trim · speed · volume · transitions · masks · text/image animations · easing curves |
| Templates | apply and extract reusable layouts · make-preset (portable text-style presets) |
| Subtitles & i18n | caption · import-srt · export-srt (line/word SRT + VTT) · translate (multi-language draft clone) |
| Effects | sfx · chroma (chroma key) · matting (smart background removal) |
| Long-form → short | cut · detect-scenes (ffmpeg scene-cut detection) · detect-silence · detect-retakes (repeated takes) |
| Automation | serve (stateless JSONL runner) · migrate · doctor · sync-timelines (8.7 mirror repair) |
capcut describe --compact lists command names, summaries, usage, and write status. Fetch a complete contract with capcut describe --command compile; repeat --command for several commands. Plain capcut describe still emits the full registry.
Full reference for every command, option, and exit code: docs/command-reference.md (简体中文: docs/command-reference.zh-CN.md).
Sponsor
capcut-cli is MIT and free forever. Sponsoring funds faster releases and same-week support for new CapCut / JianYing versions — and unlocks power-user extras:
- $5/mo · Supporter — sponsors-only release notes plus your name in
BACKERS.md. Keep the project moving. - $25/mo · Pro — invite to the private
capcut-cli-prorepo: premium template and caption-style packs, the full Claude viral-shorts pipeline, ready-to-runcompilespecs, and early-access builds. Plus priority issue triage. - $100/mo · Team — everything in Pro for up to 5 teammates, written commercial-use confirmation, your logo in this README, and priority fast-tracking of the features your team needs.
Using capcut-cli at work? The Team tier pays for itself the first afternoon it saves your engineers.
Commercial work
Separate from sponsorship: if you are building capcut-cli into a product, or you need it to do something it does not do yet, I take on a small number of integration engagements. Write to rene@renezander.com and say what you are building.
How it works
CapCut/JianYing store each project as local JSON. capcut-cli loads that store, validates against a version-aware schema, applies your edit, and writes it back atomically (with a .bak). No project files are uploaded anywhere; nothing runs as a service. See docs/version-support.md for the CapCut/JianYing versions and schema flags it understands.
Docs & examples
- docs/command-reference.md — every command and flag (简体中文)
- docs/quickstart.zh-CN.md — 剪映快速上手 (JianYing-first quickstart, Simplified Chinese)
- examples/ — end-to-end recipes (VO alignment, serve automation, batch subtitle correction)
- docs/version-support.md · docs/jianying-encryption.md
- CHANGELOG.md · Releases — what's new
- draftcat — sibling project: governed AI pipelines (Go, MIT), same single-binary, no-API design
- skillgate — sibling project: deterministic finish-line gates (Node, MIT), for when your agent reports "done" before the tests pass
Trademarks
CapCut™ and JianYing™ (剪映) are trademarks of ByteDance Ltd. This project is unofficial and is not affiliated with or endorsed by ByteDance; the marks are used nominatively to describe interoperability.
License
MIT
相關資源

MoneyPrinterTurbo Video Generation
SkillMoneyPrinterTurbo 官方技能:给一个主题或一段文案,agent 自动安装配置并生成带配音、素材、字幕和背景音乐的短视频,默认输出中文竖屏 MP4。需要一个大模型 API Key(Kimi、DeepSeek、OpenAI 等均可)和免费的 Pexels 素材 Key,支持 macOS 与 Windows。
harry0703/MoneyPrinterTurbo132k
pyVideoTrans
CLI视频翻译与配音一站式工具:串起语音识别、字幕翻译、多角色 AI 配音与音画合成,可接 Faster-Whisper、DeepSeek、Edge-TTS 等本地或在线模型;cli.py 支持无界面批处理。需要 FFmpeg。
jianchang512/pyvideotrans19.4k
VideoCaptioner
CLI卡卡字幕助手的命令行版:把课程、讲座、访谈视频转成字幕,再用大模型断句校对、翻译并烧录回视频。必剪识别与必应翻译免费可用,大模型优化需自备 API Key,需 Python 3.10–3.12。
WEIFENG2333/VideoCaptioner16.3k
Easel
CLI浙大、北大实验室开源的社媒运营助手,基于 OpenClaw,含 113 个技能,覆盖热点发现、选题排期、图文与视频创作、发布到小红书/抖音/B 站/公众号等七个平台和数据复盘;easel skill 命令可单独调用某个技能,另有 Web 工作台。需 Python 3.10+、Node.js 22.19+、FFmpeg 和大模型 API Key,发布要登录平台账号;小红书自动发布有风控风险,建议人工确认后再发。
ZJU-REAL/Easel3.7k
VectCutAPI
MCP开源的视频剪辑 API 与 MCP 服务:添加视频、音频、字幕、贴纸、特效与关键帧动画,生成可导入剪映或 CapCut 继续编辑的草稿。需本地部署 Python 服务并安装 FFmpeg。
sun-guannan/VectCutAPI2.3k
YouTube Clipper Skill
Skill下载 YouTube 视频与字幕,由 AI 按内容划分几分钟粒度的语义章节,选段后精确剪辑、翻译成中英双语字幕并烧录,同时生成小红书、抖音等平台文案。依赖 yt-dlp 与带 libass 的 FFmpeg。
op7418/Youtube-clipper-skill2.2k