aiboost

5 ms·

1. 1.▲
   by aiboost 3mo ago

   Exactly, a reproducible research workbench should itself be reproducible. Everything here is open source and MIT license. The model may still be a black box, of course you can use an open model if you prefer.
2. 2.▲
   Show HN: Open Science, open-source alternative to Claude Science(github.com)

   8 points by aiboost 3mo ago|2 comments
3. 3.▲
   by aiboost 6mo ago

   Solid Go codebase, especially the lock-free CAS rate limiting and explicit crypto.ZeroKey memory wiping. Curious how you handle token billing if an upstream SSE stream drops before the final usage chunk arrives?
4. 4.▲
   by aiboost 6mo ago

   I currently use the built-in caffeinate command in the terminal for this exact use case (keeping AI agents or large downloads running while away). However, caffeinate fails when I physically close the MacBook lid (unless connected to an ext
5. 5.▲
   by aiboost 6mo ago

   Text files + Git > Vector DBs. Nice work. I'm curious about how this scales. As the notes/ directory grows over weeks or months, reading past daily logs will eat up the context window.
6. 6.▲
   by aiboost 1y ago

   Thanks for checking the list — I’m maintaining this repo and will update it based on feedback. Quick notes: 1. This is a curated index, not runnable code — entries have one-line descriptions and tags to help scanning. 2. If something’s miss
7. 7.▲
   Show HN: AI for Science – Curated Resources(github.com)

   2 points by aiboost 1y ago|1 comments
8. 8.▲
   by aiboost 1y ago

   Hi, I'm the maintainer of Awesome-A2A. Why: Google’s Agent2Agent protocol is still new and resources are scattered. What's inside: 40+ SDKs across Go/Rust/Python/TS, demo apps, validation tools. Gaps: Looking for pr
9. 9.▲
   Show HN: Awesome-A2A – curated resources for Google's Agent2Agent protocol(github.com)

   5 points by aiboost 1y ago|1 comments