Husband & dad from @UofSC. Post own ideas on current events impacting me, my family, my community, and my country. Includes a side of F1 & college football.
@thsottiaux I'd say it's an 8/10 experience. Voice has been amazing, but it's frustrating at times. I wish there was a more intuitive way to tag and reference other conversations that were happening rather than just pasting the session ID.
@Jakubantalik@a_brinza This looks great, I made a Swift native implementation from your project as inspiration but kept you as the original credits. I look forward to pulling this into a few upcoming projects, thanks for the MIT license!
https://t.co/EfkKQaUfl2
@athyuttamre Productivity comes to an absolute halt when the voice can't tell me it's waiting for me to click approve on a computer use request. Voice just keeps saying it will look into that
@thsottiaux Have been using Voice and it seems like I found a dead end UX where asking for it to do something doesn't tell you that it's waiting on access approval. I asked it to do something and it kept telling me "Checking" but on screen needed approval for Computer Use.
@unclebobmartin@ori_pomerantz You should try out Keystone, an open source framework to capture the requirements of what you want without coupling tests to the code. I've had a lot of success and it sounds it could be a good tool to have in your toolbox.
https://t.co/01oxOURh17
@jxnlco How does everyone become developers and manage feature development?
I've published Keystone, an open source framework to help bring structure to requirements.
https://t.co/01oxOURh17
@benthompson I created Keystone to help keep Sol and other agents on the right track. Define your features as a spec with hierarchical rules that get inherited. Would love to get some feedback on this project to help structure long-term guardrails for agents.
https://t.co/LdzskEkQyV
@GergelyOrosz It's a combination of combining the apps and also making the less-technical aspects of Codex (git, commits, etc) more accessible to users via ChatGPT Work. I've had regular chat conversations offer to complete the task as a ChatGPT work task.
@LLMJunky@christcast Would it work to put in your agents.MD file that anytime they're working on a spec file, they update the top of the file to indicate their session ID? The intent being that agents auto discover tasks touching the same documents
@thsottiaux Sometimes I lose context of which project the thread is in when I come back to a thread. I can usually tell by looking at my last message but sometimes the Codex response can leave me a little confused. Visually showing a tag of the project that a thread is in could be nice.
@thsottiaux I had great plans to burn down my usage and start my weekend of coding with a fresh reset that expired today. But then i logged in and saw my reset expired before the end of day.
And now I'm at 4% thanks to some fast mode prompts I sent while getting dinner ready
@dwhitedesign This is great! I took some inspo and shipped a SwiftPackage to bring this to native swift. Kept you as the original work this is based off of and added haptic to match each sound - https://t.co/kRdLkc8ans
@pvncher 'Expires 7/17' is what date/time stamp?
Having this clarification would be super helpful to plan around since I still have a decent amount of my current weekly usage remaining.
@Dimillian If the idea works, and I stick with it, then I will want to be able to read it and maintain it. Agents tend to put logic in view files because it works but I'll often have it refactor itself and move to the ViewModel, Models before defining arch via AGENTS.md files