AiToEarn: an open-source agent that publishes to 14 platforms

Posting one video to Douyin, Xiaohongshu, TikTok and YouTube means logging in four times, uploading four times, and then babysitting four comment sections. AiToEarn (yikart/AiToEarn, TypeScript, MIT — about 26.5k stars as of 2026-09-29), the open-source project the WeChat Channels account AI纪年 recommended this August, splits that chain into four agent modules: Create, Publish, Engage and Monetize. An agent calls video and image models to draft the content, a shared calendar schedules it to 14 platforms, a browser extension works the comment sections, and brand deals settle by results. There’s a hosted web app, an MCP endpoint for Claude or Cursor, and a Docker compose file for self-hosting.

First you connect your platform accounts. The connect dialog lists all 14 channels, both Chinese and international; Xiaohongshu goes through a browser extension:

Core features

• One-click distribution to 14 platforms: Douyin, Kuaishou, Bilibili, Xiaohongshu, WeChat Channels, WeChat Official Accounts, TikTok, YouTube, Facebook, Instagram, Threads, X (Twitter), Pinterest and LinkedIn, all planned on one calendar like a content schedule.
• Agent-generated content: describe what you need and the agent calls video models (Grok, Veo, Seedance) and image models (Nano Banana) to take an idea to a finished draft; tasks can be batched and generated in parallel for running account matrices.
• Automated engagement: a browser extension likes, favorites and follows on the connected platforms, drafts a targeted AI reply to each comment, flags high-intent comments like “link please” or “how do I buy”, and monitors brand mentions.
• Brand-deal monetization: a content marketplace hands out merchant promotion tasks settled by CPS (per sale), CPE (per engagement) or CPM (per view). Task prices in the README’s screenshot range from ¥0.10 per action to ¥20 per article:

• MCP access: any MCP-compatible assistant can drive it — Claude Desktop and Cursor each have a copy-paste config, over HTTP or SSE. The 2.1 release (March 2026) added MCP plus the content marketplace and OpenClaw support, so OpenClaw users can pick up paid tasks directly.

Typical use cases

• Account matrices: batch-generate dozens of video and image posts, fan them out across accounts and platforms in parallel, and schedule everything on one calendar.
• Cross-border teams: cover Chinese channels (Douyin, Xiaohongshu, WeChat) and international ones (TikTok, Instagram, YouTube, LinkedIn) from one workflow; the 2.4 release adds per-platform limits and caption prompts to draft generation.
• Local businesses: restaurants, shops, guesthouses and gyms post promotion tasks and let creators do the online reach — a vertical added in the 1.8 release.

Quick start

Register on the official site and create a key under Settings → API Key; everything except the plain web app needs it. The China build (aitoearn.cn) and the international build (aitoearn.ai) issue separate keys — mixing them returns a 401.

Route	Best for	Deployment
Web app	Everyone	None
MCP	Claude, Cursor users	None
OpenClaw plugin	OpenClaw users	None
Docker	Self-hosting teams	A server
From source	Developers	Dev environment

MCP config for Claude Desktop (international endpoint; for the China build swap the domain to aitoearn.cn, SSE lives at /api/unified/sse):

{
  "mcpServers": {
    "aitoearn": {
      "type": "http",
      "url": "https://aitoearn.ai/api/unified/mcp",
      "headers": { "x-api-key": "YOUR_API_KEY" }
    }
  }
}

To self-host, three commands stand up the whole stack, database included:

git clone https://github.com/yikart/AiToEarn.git
cd AiToEarn
docker compose up -d

Then open http://localhost:8080. To actually publish from a self-hosted instance you point Server Relay at the official aitoearn.ai in the configuration panel, borrowing its OAuth credentials instead of registering as a developer on each platform; AI models work the same way, through AI Relay or your own keys. OpenClaw users install the plugin with npx -y @aitoearn/openclaw-plugin-cli instead.

Summary

AiToEarn fits one-person companies and content teams running accounts across several platforms who want the whole chain from drafting to monetization automated; if you only need a lightweight uploader for finished videos, social-auto-upload (Python plus browser automation, publishing only) is the smaller tool. It’s MIT-licensed TypeScript, about 26.5k stars as of 2026-09-29, and actively maintained (last push 2026-09-18).

Three caveats: a self-hosted instance either borrows the official Relay credentials or registers as a developer on each of the 14 platforms — the first ties you to their service, the second is real work; auto-liking and auto-following sit in the gray zone of most platforms’ rules, so accounts carry some risk; and generation runs on cloud models, so Docker saves you the setup, not the model bill.