A Chinese developer blogger set up OpenClaw to mine YouTube for topics every Monday, draft X threads and LinkedIn posts, and wait for approval in Telegram. He says daily content time fell from about three hours to about thirty minutes.
- First reported
- eastondev.com · 27 Feb 2026
- Added
- 13 Oct 2026
- Source checked
- 13 Oct 2026
What they did
He installed a YouTube skill to watch chosen channels and keywords and scheduled a topic research run every Monday at 9am. He iterated the drafting prompts for X threads and LinkedIn long posts more than ten times, connected both platforms through their APIs with keys kept in environment variables, and added a Telegram review step so nothing posts without his confirmation. A weekly job summarizes performance into a report.
What happened
The poster says his daily content time dropped from about three hours to about thirty minutes, and that he edits roughly 20% of each AI draft instead of writing from scratch.
Keep in mind
Early hooks were weak, drafts often ran over length limits and the AI sometimes missed the source's main point; he rates the output about 70 out of 100 and keeps a human check. No follower or engagement numbers are shared.
Try it with your agent
Every Monday at <time>, check new videos from <YouTube channels> about <topic keywords> and pick the three most useful ideas. For each, draft one X thread under <character limit> per post and one LinkedIn post, each with a strong first line. Send the drafts to me in <chat app> and publish only the ones I approve.
Replace the <angle brackets> with your own details.
Original source
eastondev.comUser storySomeone who uses the agent described their own setup. Figures are theirs.
