by dreammis · Agent Tool · ★ 15.7k
Last updated: · Indexed by AgentSkillsHub · Auto-synced every 8h
npx skills add dreammis/social-auto-upload --skill douyin-upload
Taken from the project's README on GitHub.
| Stars | 15,725 |
| Forks | 2,664 |
| Language | Python |
| Category | Agent Tool |
| License | MIT |
| Quality Score | 62/100 |
| Open Issues | 80 |
| Last Updated | 2026-10-22 |
| Created | 2023-12-04 |
| Platforms | python |
| Est. Tokens | ~17k |
social-auto-upload 是一个面向内容创作者和运营者的多平台发布工具,提供统一 CLI 和供 Codex、OpenClaw、Claude Code 等 Agent 使用的 Skill。 支持的视频、文章、图文发布平台及对应入口见下表。 CLI 命令按平台维护在 ,发布流程位于 。日常变更记录见 CHANGELOG.md,稳定版本通过 GitHub Release 发布。 💎 赞助商 感谢 DolOffer 对本项目的支持!对于做内容矩阵、多平台分发和 AI 自动化运营的创作者来说,ChatGPT、Claude、YouTube Premium、Spotify、Apple Music、Notion、Office 等数字工具往往是长期成本。DolOffer 提供 AI、视频、音乐和效率工具相关的订阅与充值服务,帮助用户更低成本地配置常用数字产品。更多说明可查看 DolOffer...
These tools work well together with social-auto-upload for enhanced workflows:
Looking for a social-auto-upload alternative? If you're comparing social-auto-upload with other agent tool tools, these 6 projects are the closest alternatives on Agent Skills Hub — ranked by topic overlap, star count, and community traction.
AutoClip|一个链接,一键出片。开源 AI 视频剪辑桌面工具,将播客、访谈、课程等长视频自动剪成短视频,生成字幕、封面和发布文案,适配抖音、小红书、TikTok、Reels 与 YouTube Shorts。Ope
AI agent skill that researches any topic across Reddit, X, YouTube, HN, Polymarket, and the web - then synthes
A curated list of awesome Claude Skills, resources, and tools for customizing Claude AI workflows
Agent skill that removes signs of AI-generated writing from text
A hand-picked collection of the finest of resources for the most awesome of agents, Claude Code, the undispute
Marketing skills for Claude Code and AI agents. CRO, copywriting, SEO, analytics, and growth engineering.
Explore other popular agent tool tools:
Grades come from a rule-based scan built on the SlowMist agent-security taxonomy: 42 published rules, covering credential harvesting, data exfiltration, curl | sh installers and more. It is a first-layer scan, not a manual audit — we say so rather than overstate it.
The scale of the problem is documented independently: Liu et al. (2026), in a study of 31,132 agent skills, report that 26.1% contain security vulnerabilities. Our own full-catalog census is published as a citable open dataset.
Sources & who's responsible: