Maintaining a reading list with claude
July 31, 2026 · Jerome Gill
I made a micro-app for tracking links Claude tells me to read.
When working in unfamiliar tech I got into the habit of getting claude to find links to justify basically every decision I wasn’t sure about.
I started keeping a .md file of links managed by a skill that would add things to it as it finds things. But it got annoying to keep on top of. I’d forget to mark things as read, and generally it was annoying.
So I vibed up a little app to manage it for me. The backend is a Claude Code skill that appends links to a YAML file. The frontend is a small Mac app made with tauri that reads that file and shows you what’s unread.
Claude never touches the read list. It only adds to the top of unread, and never adds the same link twice.
Each link gets one line on what it answers, not what it is. “When CommandBus and CommandHandler earn their keep versus a plain service” is useful later. “The NestJS CQRS docs” is not. You forget the conversation. The line has to work without it.
Click a link to open it. Click × to mark it read.
Nothing leaves your machine. The file is ~/.claude/reading-list.yaml, plain text, on your Mac.
I’ve used it a lot while planning features, especially learning new tech. I have had to do a lot of that lately, now that I’m in a role where I have much less say over tech choices.
It should be pretty helpful though. You could use it to learn basically anything Claude can point you at.
Install is three steps, about five minutes. Get the app from Releases, or build it yourself from the repo then in Claude Code:
/plugin marketplace add JeromeGill/reading-list
/plugin install reading-list@jeromegill-plugins
/reload-plugins
Then tell Claude to actually use it — add a line to ~/.claude/CLAUDE.md saying so, because Claude only uses a skill when it decides it’s relevant, and left alone that’s inconsistent.
Skill and README are both on GitHub:
Maybe it’s useful for someone else too.