readable-thrift VS promnesia

https://github.com/nccgroup/readable-thrift

Cap'N'Proto seems to have a `capnp` tool for encoding/decoding text representations to binary which seems to be officially supported and documented!

In term of automatically saving everything, There is heyday.xyz, polished but quite expensive. Or https://github.com/karlicoss/promnesia, a more experimental take.

Promnesia & theconversation.social were on similar themes/solutions.

1. you often don't know what resources you will really "value" in the future, so no more to save or not to save, this is the question

2. tagging, to be effective, require discipline (thinking about then sticking to an agile system). So, we just replace it with search, preferably NLP/AI (so you don't have to remember the exact keywords)

Apps do exist, from the expansive [1] to the experimental [2].

Personally I invested time in my filling system, and over-saving does not cause me much angst, so I’m OK with it. I also use maintenance as an occasion for renewed discovery.

[1] https://heyday.xyz/

[2] https://github.com/karlicoss/promnesia

I want to build a browser that captures everything I saw on the internet, allows me to search it, run graph algorithms (like PageRank). Improves navigation (by showing trails as tree instead of tabs). Heavily offline focused (Backend only for updates, maybe for analytics).

Difference with rewind.ai: linkkraft does not have funding, i'm solo, no apps & image/video/audio recognition. Focus on web, trails, research and using web copies, selections/highlights as part of your notes & whiteboards. Preserving all possible graphs.

My inspirations: https://pages.gseis.ucla.edu/faculty/bates/berrypicking.html, https://beepb00p.xyz/promnesia.html, Jeff Raskin (Global Search, Zoom UI) https://linkkraft.com/notes/backstory

I've built a prototype with trails tree & HTML snapshoting. For each my step even inside SPA linkkraft creates HTML snapshot.

Promnesia works like that: https://github.com/karlicoss/promnesia/