Turning My Coding Streams Into Blog Posts (With a Little Help From Claude)

I’ll be honest: I’m lazy when it comes to writing blog posts. Not lazy about sharing — I try to stream regularly, push code to GitHub, drop things in the community Discord. But sitting down to write a structured article about something I just spent two hours live coding? That rarely happens.

The stream is the content. I jump on, explain what I’m doing, answer questions live, push something to GitHub, maybe add chapters via AI later. Done. No editing, no polishing.

The problem is that not everyone wants to watch a two-hour live stream. Some people prefer reading. And if you want to skim for the specific part you need, a video is the worst format possible.

So I finally did something about it.

The Pipeline

I built a small Claude Code skill — /my-video-to-blog — that takes a YouTube URL and does the following:

  1. Downloads the auto-generated captions from YouTube (no re-transcription needed if they exist)
  2. Downloads the video and extracts ~50 frames at regular intervals, with timestamps burned in
  3. Claude reads the transcript, reviews all the frames, picks the most useful screenshots, and writes a structured blog post in my writing style
  4. As a bonus pass, it searches official docs for anything I mentioned but didn’t explain, and adds clearly marked callout blocks with links

The whole thing runs in one command per video — or in a batch. To clear the back catalogue, I had Claude run through all 26 streams sequentially, one sub-agent per video, committing each post to the repo as it finished. Total wall time: 2h 34m 42s.

Claude Code session showing “Crunched for 2h 34m 42s”

I came back, reviewed the drafts, fixed a couple of things, and pushed.

It’s not perfect — the transcript-based posts lack the back-and-forth of a live stream, and Claude occasionally fills gaps with context I’d phrase differently. But that’s what the review step is for.

The video stays on YouTube. The blog post is the distilled version. Both exist, both link to each other.

Why This Also Helps the BC Community AI Tools

There’s a side effect worth mentioning. The community BC intelligence tool CentralQ — built by Dmitry Katson — scrapes blog posts and documentation to answer BC-related questions. It works well for written content but has trouble with YouTube live streams specifically.

So every blog post generated from a stream is also an improvement to that tool’s knowledge base. Two birds, one pipeline.

All the Generated Posts

26 posts, covering streams from May 2024 to February 2026. The full back catalogue.