This is a cool project! I usually review docs/blogs/etc. as much as possible and brainstorm with an llm. I research to see if there are any other open source examples that I can also learn from. I find that this does pretty well. At the end of the day, after the implementation, I continue testing and that's where it gets interesting since I'll run into edge cases that creators of the original product already ran into and resolved and over time it helps make the software more robust as I use it and fix it myself.
I'll check this out. I actually recently was looking for smth like this to be able to track my emotions daily because I think that otherwise I'm not as self-aware. What results have you been seeing from logging and why this diagram structure? And one other note is that I liked that I was able to log my mood initially to test without having to login, reduces friction.
This looks really good. NixOS is a great operating system for reproducibility. I like also that you can spin up VMs via the CLI just as a convenience part too. This looks great for an individual developer. From the team perspective, if I create a golden snapshot of a dev env, is there a way within the product to sharae the image with others so they can spin it up instantly?
This is great, I've tried out automated podcast editing tools before and they cut too aggressively in my experience. What are you thinking about doing next with this now that you've gotten the alignment snapping working cleanly for 'um' and 'ah', are you thinking of expanding the tool?
This is a really interesting benchmark and also timely given a lot of existing benchmarks don't do a good job. The mechanics and edge cases seem notoriously difficult to parse also even for perhaps human players. Have you been also plugging these into newer reasoning models to see how providing them with thinking time improves their win rate against the baseline?
This is super cool. I've been a heavy tmux user for a long time and using it more with my coding agent sessions and prefer ghostty. What was the biggest challenge when it came to building a multiplexer directly on top of libghostty that you ran into?
This is super cool. I like the anti magic and explicit philosophy you have. Do you plan on eventually building out the advanced optimization passes so that the language is faster?
Very cool to see more security focused tools being built here for the Nix ecosystem. What were some of the biggest roadblocks or challenges you hit when building this?
This is interesting way of presenting content and I do feel like engaging educational content is a valuable thing to work on. For the historical figures, how are you handling grounding the facts/reducing hallucinations that may conflict with historical accuracy? Also do you plan on making it possible for people to add their own historical figures like a registry of sorts?
This is cool, I'll check it out. Do you guys think that there will still be some work that is done locally vs the cloud, how are you thinking about the split from what you've seen developers doing? I also agree with the fact that git worktrees can be super frustrating to manage especially across multiple agent sessions.
This is an interesting approach. I think people still underestimate how quickly token limits and context bloat become the bottleneck when you start running agents in production loops. Seems very relevant topic also given the environment. Will check it out
Also an eng in big tech and there are constantly new initiatives I think we get asked to think about or do but main value I've found is by solving for my own problems with AI.
Specifically, I built out a workflow layer for my coding agents. One key workflow I use is an end-to-end pipeline that starts with deep researching the codebase, feeds into my custom ralph or goal implementation depending on the scope, and follows through all the way to automated PR review. The upfront engineering investment to get it working reliably was heavy, but the ROI on my daily velocity has been absolutely worth it. But it took a lot of iteration to not generate slop and burn tokens.
Beyond just the productivity gains, the process of building it from scratch was actually the biggest win. It forced me to get hands-on and learn the critical, pragmatic nuances of agentic engineering. I did this all outside of working hours as my own personal open source project. Other folks around me are using it. I don't know how things will shape up but part fear, part curiosity drove me to just try to learn as much as I can with so many things outside of my control.