We hoard links like dragons hoard gold. Here is how I finally cured my "Bookmark Bankruptcy".
If you think about it, 99% of the internet's value is just URLs. A groundbreaking AI paper, a brilliant Figma UI kit, a GitHub repo, or a YouTube tutorial—they all boil down to a simple string starting with https://.
Like many of you, I suffer from digital hoarding. My browser bookmarks were a black hole. "Read it later" essentially meant "Save it forever and never look at it again." I tried moving them to Notion, Obsidian, and Raindrop, but the friction of tagging and organizing always defeated me.
I realized my fundamental approach to Personal Knowledge Management (PKM) for links was wrong.
The Shift in Mindset: From "Storage" to "Dashboards" Links shouldn't be buried in hierarchical folders; they should be visualized as active, curated dashboards. But I couldn't find a tool that was both completely private (no SaaS lock-in) and smart enough to organize the mess for me.
So, I built an open-source experiment called NaviCube.
I challenged myself to build it as a Single-file App (just one HTML file). No backend, no database. Everything lives in your browser's LocalStorage.
But here is where the PKM method actually works:
The AI Librarian: I integrated a BYOK (Bring Your Own Key) setup for OpenAI/Claude. Instead of manually sorting 500 messy bookmarks, the "Smart Tidy" feature reads the context of the URLs and automatically generates a logical MECE (Mutually Exclusive, Collectively Exhaustive) category structure for you.
The "Shareable" Output: The end goal of organizing knowledge is output. NaviCube lets you export your setup as a standalone static HTML file.
I initially built this for myself, but the use cases have been wildly surprising:
Researchers are using it to index local file:/// PDF paths.
Teachers are dropping the exported HTML onto school servers as clear "Course Resource Pages."
Frontend Teams are using it as a zero-maintenance shared resource hub hosted on GitHub Pages.
I'm fascinated by the intersection of Local-first software and AI. I just open-sourced it and put it on Product Hunt.
I would love to hear from this community: How do you handle the friction of categorizing hundreds of links? Are "Single-file Apps" a viable trend for PKM tools in future?
https://github.com/xu-hao-yuan/navicube
Replies
Single-file, no backend, straight to LocalStorage is a genuinely disciplined constraint to hold yourself to most "local-first" tools quietly need a server somewhere the moment they add a feature. Respect for actually shipping under that constraint through the AI tidy feature too.
On your second question: I don't think single-file apps are the trend so much as a specific symptom of a trend people are done trusting SaaS with things they can't get back out. The format is almost incidental to that.
On the first one, and the thing I'd actually push on: does Smart Tidy solve the categorizing friction, or just make the pile look tidier? A clean MECE structure over 500 links is satisfying to look at, but the real test is six months from now when you need the one link relevant to a decision you're making and you don't remember which category you'd have filed it under at the time. Auto-sorting solves browsing. It doesn't obviously solve the resurfacing problem, which was usually the actual reason the bookmarks piled up in the first place.