Claude Code helped me turn my OSHO preservation work into an actual archive pipeline—without having to spend hours coding everything from scratch.
The collection itself is 1300+ DVDs so far, spanning over 35 years of talks, interviews, events, discourse series, and rare recordings.
One of the largest high-quality private OSHO video collections I’ve seen anywhere... nearly complete and still growing.
The goal is to preserve the archive by converting the original DVD collection into lossless MKVs.
The archive is massive, so at one point I paused the DVD ripping workflow and started uploading completed ISO backups to Amazon while it was still unlimited.
Then Amazon shut the service down.
The web interface is limited to 5GB downloads or 1000 files at a time. There is no Linux app.
Meanwhile, I had ~10TB of already-ripped ISO backups sitting there waiting to be recovered and verified.
So the engineering side became its own project.
I fired up Claude Code and vibe coded an Amazon Photos downloader script.
The downloader evolved into an interactive archive browser. I can navigate the folder tree, queue entire sections of the archive, see file counts, sizes, totals, storage requirements, and launch batch jobs.
It rebuilds the original hierarchy, resumes interrupted work, verifies every file locally, audits completed runs, and refuses to silently accept incomplete downloads.
Downloads fail.
Discs fail.
Navigation structures fail.
Manual processing doesn't scale.
The converter walks the archive automatically, finds ISOs, batch converts them with headless `makemkvcon`, preserves chapters, skips completed discs, writes the MKV back into the same structure, and logs everything.
Validation is automated too. ffmpeg fully decodes every finished file and flags corruption automatically.
The workflow now looks like this:
0. `validate_mkvs.sh` — one-time scan of pre-existing MKVs (no ISOs = validation only)
1. `download_photos.py` — download ISO backups from Amazon Photos
2. `batch_mkv.sh` — convert ISOs to MKV, validate, optionally delete ISOs
3. `recover_corrupt.sh` — auto-fix anything flagged corrupt by step 2
4. `dvd_recover.sh` — manual recovery from physical disc for what step 3 can't fix
5. Repeat steps 1–4 as more ISOs download
The archive keeps moving.
Problem discs have their own recovery path: ddrescue → bsdtar → mkvmerge.
Sometimes the navigation layer is broken.
The video isn't.
So the workflow bypasses the damaged DVD structure and rebuilds from the raw VOB streams.
The cool part is these aren’t just scripts anymore.
They’re becoming a preservation toolkit built around the realities of maintaining a massive archive.
Less babysitting.
More automation.
More preservation.
#IKITAO #OSHO
View quoted note →
View quoted note →