隆重推出 video-use — 用 Claude Code 编辑视频。100% 开源。

将原始素材放入文件夹，与 Claude Code 聊天，即可得到 final.mp4 返回。适用于任何内容——谈话、蒙太奇、教程、旅行、访谈——无需预设或菜单。

在 Browser Use Cloud 中尝试 video-use。

• 去除填充词（umm、uh、开头失误）和片段间的沉默间隔
• 自动调色 每个片段（温暖的电影调色、中性 punch 或任何自定义 ffmpeg 链）
• 每处剪切 30ms 音频淡入淡出，你永远不会听到爆音
• 烧录字幕 按你的风格——默认 2 个单词的大写块，完全可自定义
• 生成动画叠层 通过 HyperFrames、Remotion、Manim 或 PIL——在并行子代理中生成，每个动画一个
• 自评估渲染输出 在每个剪切边界处，并在展示给你之前检查
• 在 project.md 中持久化会话记忆，下周的会话可以从上次离开的地方继续

将其粘贴到 Claude Code、Codex、Hermes、Openclaw 或任何具有 shell 访问权限的代理中：

Set up https://github.com/browser-use/video-use for me.

Read install.md first to install this repo, wire up ffmpeg, register the skill with whichever agent you're running under, and set up the ElevenLabs API key — ask me to paste it when you need it. Then read SKILL.md for daily usage, and always read helpers/ because that's where the editing scripts live. After install, don't transcribe anything on your own — just tell me it's ready and wait for me to drop footage into a folder.

代理会处理克隆、依赖项、技能注册，并在需要时提示你输入 ElevenLabs API 密钥（从 elevenlabs.io/app/settings/api-keys 获取）。

然后将代理指向包含原始拍摄片段的文件夹：

cd /path/to/your/videos claude # 或 codex、hermes 等

若要通过自己的 VPS 或 Telegram 实现始终在线的编辑，可通过 Browser Use Box 运行代理。观看 15 秒演示。

在会话中：

把这些编辑成一个发布视频

它会盘点素材，提出策略方案，等你确认，然后在素材旁生成 edit/final.mp4。所有输出都位于 <videos_dir>/edit/ — 技能目录保持干净。

如果你更想手动操作：

1. 克隆仓库并软链接到代理的技能目录

git clone https://github.com/browser-use/video-use ~/Developer/video-use ln -sfn ~/Developer/video-use ~/.claude/skills/video-use # Claude Code

ln -sfn ~/Developer/video-use ~/.codex/skills/video-use # Codex

2. 安装依赖

cd ~/Developer/video-use uv sync # 或: pip install -e . brew install ffmpeg # 必要 brew install yt-dlp # 可选，用于下载在线素材

3. 添加 ElevenLabs API 密钥

cp .env.example .env $EDITOR .env # ELEVENLABS_API_KEY=...

LLM 从不观看视频。它阅读它——通过两层提供裁剪所需所有信息，达到词边界精度。

第 1 层 — 音频转录（始终加载）。 每个源调用一次 ElevenLabs Scribe，提供词级时间戳、说话人分离和音频事件（(laughter)、(applause)、(sigh)）。所有片段打包到一个约 12KB 的 takes_packed.md — LLM 的主要阅读视图。

C0103 (duration: 43.0s, 8 phrases)

[002.52-005.36] S0 Ninety percent of what a web agent does is completely wasted. [006.08-006.74] S0 We fixed this.

第 2 层 — 视觉合成（按需）。 timeline_view 可以为任意时间范围生成胶片条 + 波形 + 单词标签的 PNG 图片。仅在决策点调用 — 模糊的停顿、重新拍摄对比、剪切点正确性检查。

朴素方法：30,000 帧 × 1,500 tokens = 4500 万 tokens 噪音。Video Use：12KB 文本 + 少量 PNG。

与 browser-use 为 LLM 提供结构化 DOM 而非截图的想法相同 — 不过这次用于视频。

转录 ──> 打包 ──> LLM 推理 ──> EDL ──> 渲染 ──> 自我评估 │ └─ 有问题？修复并重新渲染（最多 3 次）

自我评估循环在每个剪切边界处对渲染输出运行 timeline_view — 捕捉视觉跳跃、音频爆音、隐藏字幕。你只在它通过后才看到预览。

1. 文本 + 按需视觉。 不转储帧。转录是表面。
2. 音频为主，视觉跟随。 剪切来自语音边界和静音间隙。
3. 询问 → 确认 → 执行 → 自我评估 → 持久化。 未经策略批准切勿触碰剪切。
4. 对内容类型零假设。 先观察，再询问，然后编辑。
5. 12 条硬性规则，其余艺术自由。 制作正确性不容妥协。品味不是。

查看 SKILL.md 获取完整的生产规则和编辑技巧。

Introducing video-use — edit videos with Claude Code. 100% open source.

Drop raw footage in a folder, chat with Claude Code, get final.mp4 back. Works for any content — talking heads, montages, tutorials, travel, interviews — without presets or menus.

Try video-use in Browser Use Cloud.

• Cuts out filler words (umm, uh, false starts) and dead space between takes
• Auto color grades every segment (warm cinematic, neutral punch, or any custom ffmpeg chain)
• 30ms audio fades at every cut so you never hear a pop
• Burns subtitles in your style — 2-word UPPERCASE chunks by default, fully customizable
• Generates animation overlays via HyperFrames, Remotion, Manim, or PIL — spawned in parallel sub-agents, one per animation
• Self-evaluates the rendered output at every cut boundary before showing you anything
• Persists session memory in project.md so next week's session picks up where you left off

Paste into Claude Code, Codex, Hermes, Openclaw, or any agent with shell access:

Set up https://github.com/browser-use/video-use for me.

Read install.md first to install this repo, wire up ffmpeg, register the skill with whichever agent you're running under, and set up the ElevenLabs API key — ask me to paste it when you need it. Then read SKILL.md for daily usage, and always read helpers/ because that's where the editing scripts live. After install, don't transcribe anything on your own — just tell me it's ready and wait for me to drop footage into a folder.

The agent handles the clone, dependencies, skill registration, and prompts you once for your ElevenLabs API key (grab one at elevenlabs.io/app/settings/api-keys).

Then point your agent at a folder of raw takes:

cd /path/to/your/videos
claude    # or codex, hermes, etc.

For always-on editing from your own VPS or Telegram, run the agent through Browser Use Box. Watch the 15-second demo.

And in the session:

edit these into a launch video

It inventories the sources, proposes a strategy, waits for your OK, then produces edit/final.mp4 next to your sources. All outputs live in <videos_dir>/edit/ — the skill directory stays clean.

If you'd rather do it by hand:

# 1. Clone and symlink into your agent's skills directory
git clone https://github.com/browser-use/video-use ~/Developer/video-use
ln -sfn ~/Developer/video-use ~/.claude/skills/video-use        # Claude Code
# ln -sfn ~/Developer/video-use ~/.codex/skills/video-use       # Codex

# 2. Install deps
cd ~/Developer/video-use
uv sync                         # or: pip install -e .
brew install ffmpeg             # required
brew install yt-dlp             # optional, for downloading online sources

# 3. Add your ElevenLabs API key
cp .env.example .env
$EDITOR .env                    # ELEVENLABS_API_KEY=...

The LLM never watches the video. It reads it — through two layers that together give it everything it needs to cut with word-boundary precision.

Layer 1 — Audio transcript (always loaded). One ElevenLabs Scribe call per source gives word-level timestamps, speaker diarization, and audio events ((laughter), (applause), (sigh)). All takes pack into a single ~12KB takes_packed.md — the LLM's primary reading view.

## C0103  (duration: 43.0s, 8 phrases)
  [002.52-005.36] S0 Ninety percent of what a web agent does is completely wasted.
  [006.08-006.74] S0 We fixed this.

Layer 2 — Visual composite (on demand). timeline_view produces a filmstrip + waveform + word labels PNG for any time range. Called only at decision points — ambiguous pauses, retake comparisons, cut-point sanity checks.

Naive approach: 30,000 frames × 1,500 tokens = 45M tokens of noise. Video Use: 12KB text + a handful of PNGs.

Same idea as browser-use giving an LLM a structured DOM instead of a screenshot — but for video.

Transcribe ──> Pack ──> LLM Reasons ──> EDL ──> Render ──> Self-Eval
                                                              │
                                                              └─ issue? fix + re-render (max 3)

The self-eval loop runs timeline_view on the rendered output at every cut boundary — catches visual jumps, audio pops, hidden subtitles. You see the preview only after it passes.

1. Text + on-demand visuals. No frame-dumping. The transcript is the surface.
2. Audio is primary, visuals follow. Cuts come from speech boundaries and silence gaps.
3. Ask → confirm → execute → self-eval → persist. Never touch the cut without strategy approval.
4. Zero assumptions about content type. Look, ask, then edit.
5. 12 hard rules, artistic freedom elsewhere. Production-correctness is non-negotiable. Taste isn't.

See SKILL.md for the full production rules and editing craft.