抖音生成短视频

根据主题自动搜索整理信息，生成多风格视频脚本，并调用数字人平台创建完整短视频。

Security Scan

Skill purpose (generate scripts and call avatar APIs) is plausible, but there are inconsistent environment-variable declarations and minor mismatches between documentation and code that you should verify before installing or supplying API keys.

Purpose & Capability

Name/description match the code: pipeline builds a script and can call HeyGen/D-ID/Synthesia to create videos. The files implement searching (stubbed), organizing, script generation, and avatar API calls — consistent with stated purpose. Minor mismatch: SKILL metadata lists no required env vars, while SKILL.md documents multiple provider keys (HEYGEN_API_KEY, DID_API_KEY, SYNTHESIA_API_KEY) — the code actually reads only AVATAR_API_KEY.

Instruction Scope

Runtime instructions and code operate within the expected scope: generate/format content, write files under an output directory, and call third‑party avatar APIs. The pipeline writes local JSON/MD/report files and does not attempt to read unrelated system paths. The search step is intentionally local/stubbed (the SKILL.md mentions fallbacks), which is implemented in code.

Install Mechanism

No install spec; this is an instruction + code bundle that runs with Node. Nothing is downloaded or extracted during install, so installation risk is low. The package uses only core Node modules (fs, https, path); no external package installs are declared.

Credentials

SKILL.md asks for multiple provider-specific API keys (AVATAR_API_KEY, HEYGEN_API_KEY, DID_API_KEY, SYNTHESIA_API_KEY), but the code only reads process.env.AVATAR_API_KEY (and the AvatarVideoGenerator expects a single apiKey value). The registry metadata claims no required env vars. This mismatch is unexplained and could lead to user confusion or accidental disclosure of multiple keys if the user follows SKILL.md rather than code. Requiring a single API key (or clearly documenting provider-specific usage) would be proportionate; the current documentation/metadata/code inconsistency is a red flag to verify before supplying secrets.

Persistence & Privilege

Skill is not always-enabled and does not request elevated/persistent agent privileges. It does not modify other skills or system configuration. It runs as a normal Node script when invoked.

What to consider before installing

This skill largely does what it says (generate scripts and call avatar/video provider APIs), but there are inconsistencies you should resolve before installing or providing API keys: - Confirm which environment variable the code actually uses: avatar-generator.js reads AVATAR_API_KEY. The SKILL.md lists HEYGEN_API_KEY, DID_API_KEY, and SYNTHESIA_API_KEY as well — that documentation may be outdated or misleading. Only give the minimum key required (preferably a provider-specific, scoped API key) and avoid pasting multiple unrelated credentials. - Because the source/homepage are unknown, review the avatar-generator.js implementation (it sends Bearer Authorization to provider endpoints) and verify the endpoints are correct for your provider. Prefer creating limited-scope API keys on the provider side. - The skill writes files under an output directory. Run it in a sandboxed directory or container if you want to limit filesystem impact. - If you rely on the skill to perform live web searches, note the current implementation uses a local stub for search (no live scraping). If you or someone adds network scraping/third-party search integrations later, re-audit those changes. - If you need higher assurance, ask the author for provenance (repository, homepage) or a signed release, or run the code in an isolated environment and inspect network traffic when generating a test job.

Content Pipeline Skill

描述

这个 Skill 提供了一个完整的内容生成 pipeline： 搜索信息 → 整理数据 → 生成脚本 → 生成数字人视频

何时使用

当用户要求：

• "帮我生成一个关于 X 的视频脚本"
• "我想要一个数字人视频讲解 X"
• "帮我整理 X 的信息并生成脚本"
• "创建一个关于 X 的内容"

功能

1. 信息搜索和整理

• 从多个来源搜索信息
• 自动提取关键点
• 结构化数据组织

2. 脚本生成

• 支持多种风格 (分析、故事、教育)
• 自动时长分配
• 多格式输出 (Markdown, JSON)

3. 数字人视频生成

• 支持 HeyGen, D-ID, Synthesia
• 自动 API 调用
• 视频下载和处理

使用示例

基础用法

node pipeline.js --topic "人工智能的未来" --style "analysis"

指定提供商

node pipeline.js --topic "气候变化" --provider "d-id" --style "educational"

自定义输出

node pipeline.js --topic "区块链" --output "./my-videos" --style "story"

输出

Pipeline 会生成以下文件：

output/[timestamp]/
├── 01-search-results.json      # 搜索结果
├── 02-organized-data.json      # 整理后的数据
├── 03-script.md                # Markdown 脚本
├── 03-script.json              # JSON 脚本
├── 04-video-result.json        # 视频生成结果
└── REPORT.json                 # 完整报告

配置

环境变量

# 数字人 API Keys
export AVATAR_API_KEY="your-api-key"
export HEYGEN_API_KEY="your-heygen-key"
export DID_API_KEY="your-d-id-key"
export SYNTHESIA_API_KEY="your-synthesia-key"

脚本模板

支持的风格：

• analysis - 深度分析型
• story - 故事型
• educational - 教育型

扩展

添加新的脚本风格

编辑 script-generator.js 中的 SCRIPT_TEMPLATES：

SCRIPT_TEMPLATES['my-style'] = {
  intro: '...',
  body: '...',
  conclusion: '...'
};

添加新的数字人提供商

编辑 avatar-generator.js 中的 generateWithProvider 方法。

集成新的搜索源

编辑 pipeline.js 中的 searchInformation 方法。

故障排除

网络搜索不可用

• 使用本地数据或用户输入
• 集成飞书知识库
• 使用 curl/wget 备选方案

数字人 API 失败

• 检查 API Key 是否正确
• 验证网络连接
• 查看 API 文档

脚本质量不佳

• 调整脚本模板
• 改进数据整理逻辑
• 添加 AI 内容增强

文件结构

content-pipeline/
├── script-generator.js         # 脚本生成引擎
├── data-organizer.js           # 数据整理工具
├── avatar-generator.js         # 数字人视频生成
├── pipeline.js                 # 主 pipeline
├── SKILL.md                    # 本文件
└── examples/
    ├── sample-data.json        # 示例数据
    └── sample-script.json      # 示例脚本

许可证

MIT

支持

如有问题，请查看日志文件或联系开发者。