There is a skill-making skill that comes with Claude Desktop. You can import it to Claude Code, or find it on Github.

In just the last hour I've had Claude Code teach itself how to use both the Logseq CLI and the Zotero local API and save that knowledge as a "skill" so it can be referenced whenever I need it!

Interesting to compare @maximillianalvarez.bsky.social's excellent reporting with this dismissive and condescending post by @gruber.foo back in April.

Archived version: archive.ph/UREGx

But I found a solution! A command line app that runs off of Tesseract OCR and supports over 100 languages. Like ABBYY it can do two languages at once, and it can also compress PDFs to take up less space. You can find it here:

github.com/ocrmypdf/OCR...

For instance, you can show all pages with a given tag, or all pages in a particular folder path. Then you can check off which properties to display in the table by checking them off in the property menu.

I'm loving it, but I agree. Hopefully after they iron out all the bugs and launch sync, etc. they can focus on usability - especially on mobile. Sync will be part of Logseq Pro (along with collaboration and publishing) which will hopefully give them the money they need to hire someone for this.

There are a couple available for Obsidian already. Eventually, should be possible for someone to make one to work with Logseq DB...