Skip to content
Back to skills

Videocut 0xfango

BSecurity

环境准备。安装依赖、下载模型。触发词:安装、环境准备、初始化

  • 57 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added October 17, 2026
toolspythonbashnodenodejs

Security analysis

B80/100
  • criticalModifies startup scripts or system services for persistence
  • mediumInstalls packages at runtime which could introduce malicious dependencies

Pro shows the line behind each finding and how to fix it

Scanned October 17, 2026

npx -y skills add diegosouzapw/awesome-omni-skill --skill videocut-0xfango --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Videocut 0xfango?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Videocut 0xfango
[![Security: B — Skills Directory](https://www.skillsdirectory.com/api/skills/diegosouzapw-videocut-0xfango/badge)](https://www.skillsdirectory.com/skills/diegosouzapw-videocut-0xfango)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: videocut:安装
description: 环境准备。安装依赖、下载模型。触发词:安装、环境准备、初始化
---

<!--
input: 无
output: 环境就绪
pos: 前置 skill,首次使用前运行

架构守护者:一旦我被修改,请同步更新:
1. ../README.md 的 Skill 清单
2. /CLAUDE.md 路由表
-->

# 安装

> 首次使用前的环境准备(本地模式)

## 快速使用

```
用户: 安装环境
用户: 初始化
```

### 依赖清单

| 依赖 | 用途 | 安装命令 |
|------|------|----------|
| Python 3.8+ | 运行 FunASR | `brew install python` |
| funasr | 语音识别 | `pip install funasr` |
| modelscope | 模型下载 | `pip install modelscope` |
| FFmpeg | 视频处理 | `brew install ffmpeg` |
| Node.js 18+ | 运行转录模块 | `brew install node` |

### 模型清单

首次运行自动下载到 `~/.cache/modelscope/`:

| 模型 | 大小 | 用途 |
|------|------|------|
| paraformer-zh | 953MB | 语音识别(带时间戳) |
| punc_ct | 1.1GB | 标点预测 |
| fsmn-vad | 4MB | 语音活动检测 |
| **小计** | **~2GB** | |

### 安装步骤

#### 1. 安装系统依赖

```bash
# macOS
brew install python node ffmpeg

# Ubuntu
sudo apt install python3 python3-pip nodejs ffmpeg
```

#### 2. 安装 Python 依赖

```bash
pip install funasr modelscope
```

#### 3. 下载模型(约2GB)

```bash
cd /path/to/videocut-skills/安装/scripts

# 自动下载所有模型
python test_funasr_local.py --download
```

#### 4. 验证环境

```bash
cd /path/to/videocut-skills/安装/scripts

# 快速验证(检查依赖)
python test_funasr_local.py

# 综合验证(加载完整模型)
python test_funasr_local.py --verify
```

成功输出:
```
🎉 本地模式完全就绪!可以使用以下命令转录:

   python 剪口播/scripts/transcribe_local.py video.mp4
```

---

## 完整安装流程

```
1. 安装系统依赖(Python、Node.js、FFmpeg)
       ↓
2. 安装 Python 依赖(funasr、modelscope)
       ↓
3. 下载模型(约2GB)
       ↓
4. 验证环境
       ↓
5. 完成 ✅
```

---

## 常见问题

### Q1: 模型下载慢

**解决**:使用国内镜像或手动下载

### Q2: ffmpeg 命令找不到

**解决**:确认已安装并添加到 PATH

```bash
which ffmpeg  # 应该输出路径
```

### Q3: Node.js 版本太低

**解决**:需要 Node.js 18+

```bash
node --version  # 需要 v18.x 或更高
```

### Q4: pip 和 python 版本不一致

**解决**:确保 pip 和 python 指向同一个版本

```bash
# 检查版本
python --version
pip --version

# 如果不一致,在 ~/.zshrc 添加 alias
alias python=python3.11
alias pip=pip3.11
```

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.