Automation#Social media automation#Browser automation

social-auto-upload: one-click multi-platform video distribution

Open-source Python tool that publishes one video to eleven platforms including Douyin, Bilibili, and YouTube via browser automation; ten ship a unified CLI.

Project facts

GitHub Ecosystem
Repositorygithub.com/dreammis/social-auto-upload
License
MIT
Language
Python
Stars
15,399
Data checked
2026-10-11

Snapshot figures reflect the check date and may change over time.

Publishing the same short video to Douyin, Bilibili, and Xiaohongshu usually means logging in, typing a title, and uploading a file three separate times. social-auto-upload hands that process to a script: written in Python, it drives each platform’s web frontend with browser automation and relies on no official API. After a dense refactor in August 2026, ten of the eleven supported platforms run through one unified sau CLI — YouTube got its CLI too — and the repo ships a bootstrap prompt that hands installation and platform verification to coding agents like Claude Code and Codex. The project counts about 15,400 GitHub stars and 2,600 forks (checked October 2026); GitHubDaily recommended it in April 2025.

TikTok upload demo: the program fills in the title, sets a schedule, and completes the publish on its own

Core features

  • Eleven platforms, ten CLIs: Douyin, Kuaishou, Xiaohongshu (RedNote), Bilibili, WeChat Channels, Weibo, Baijiahao, Alipay Life Account, and Hupu on the domestic side, plus TikTok and YouTube. Every platform except TikTok — which still relies on the legacy Chrome example script — is wired into the unified sau CLI.
  • YouTube without the API lockdown: uploads run through browser automation, and the README is blunt about why — videos uploaded through the official API by projects that haven’t passed Google’s compliance audit get locked to private with no way to make them public. The CLI supports --playlist for serialized series and --visibility (public, unlisted, or private), waits for the upload to reach 100% before hitting publish, and honors a YT_PROXY setting in conf.py for regions where YouTube is unreachable.
  • Image posts and scheduling: Douyin, Xiaohongshu, and Kuaishou accept image-and-text notes; Douyin, Bilibili, Xiaohongshu, Kuaishou, WeChat Channels, and TikTok support scheduled publishing. Scheduling is computed as “the next day,” a reflection of the author’s own queue-a-day-ahead habit — immediate posting means adjusting that yourself.
  • A bootstrap prompt for agent clients: docs/agent-bootstrap.md ships a copy-paste prompt; hand the repo plus that prompt to OpenClaw, Codex, or Claude Code and the agent installs the project on the uv + sau mainline, then verifies the entry points of the four accepted platforms: bilibili, douyin, kuaishou, and xiaohongshu.
  • Accounts and verification codes: each account_name maps to one account file, and tasks can run concurrently per account. When Douyin triggers SMS verification, the program reads verify_code.txt from the project root, or you can type the code straight into the terminal.
  • Bilibili without manual setup: the CLI downloads biliup on first run and keeps it updated from upstream; if the terminal QR code renders badly, scan qrcode.png from the working directory instead.
  • Platform quirks worth knowing: Weibo caps titles at 30 characters and Hupu at 4–40; some platforms lack scheduling or image-post support. The README’s capability matrix spells out account setup, CLI, and skill support for each platform, and it’s worth a look before you commit to one.

Typical use cases

  • Batch-distribute a finished video with sau, skipping the platform-by-platform upload grind.
  • Matrix operation: give each account its own config and run each account’s scheduled publishing tasks in batch.
  • Delegate the whole install-login-upload pipeline to an AI agent and spend your time making content.

Quick start

Clone the repo and set up a virtual environment with uv (see docs/install.md in the repo for details):

git clone https://github.com/dreammis/social-auto-upload.git && cd social-auto-upload
uv venv && uv pip install -e . && patchright install chromium
cp conf.example.py conf.py

Log in to an account first, then run an upload:

sau douyin login --account <account_name>
sau douyin upload-video --account <account_name> --file videos/demo.mp4 --title "Sample title" --desc "Sample description"

If you’d rather have an agent set things up, send the repo together with the bootstrap prompt from docs/agent-bootstrap.md. For a GUI-first alternative in the same space, see AiToEarn, which covers 14 platforms with built-in AI generation; this project’s edge is a fully local Python CLI with first-class agent support.

Summary

This suits short-video creators, multi-account matrix operators, and independent developers who want uploads scripted and agent-driven. It’s MIT licensed and fine for closed-source commercial use. The trade-offs are real: browser automation means a real account with QR-code or cookie login on every platform, and when a platform redesigns its upload page the matching uploader breaks — the August refactor was largely chasing a WeChat Channels redesign and Tencent’s real-name verification flow. Refactoring continues, headless mode on the patchright driver is the current mainline, and the old web UI is unmaintained; the official docs admit they lag behind the code, so when something won’t install, read the in-repo docs/ first or hand the whole thing to an agent.