Hey friend. Can you spare a ⭐️ for a (hopefully) interesting open source project I'm working on? github.com/ExpeditedProjects…

Every Monday I ask Claude to make a 1 page PDF of my AI usage, new skills, what was unused and what to try in a futile attempt to keep up with the AI dev tooling changes that are happening.

What's best practice for organizing agents (Grokbot, Hermes, Openclaw?) One per task or one per role?

Find out which members of Congress actually voted for the issues you care about at dollarvote.us It's a site I put together for Hacktoberfest where local devs came together for a day of hanging out and creating cool stuff. It uses Google's Gemma AI to analyze in detail what each bill from the last congressional session was about and then how individual legislators voted on them.

Entering the brave new world of hardware programming (as soon as my NFC reader + cards get here).

What's the term for a "fake applicant" who is applying for a role? I keep hearing stuff like: "The person we hired is not who was doing the work", "The person we interviewed had stolen a person's resume and was pretending to be them", but I don't have a succint term to wrap up this concept. Maybe "applicant fraud"?

Claude, Codex, whatever, go ask it: "Am I actually using all these skills and plugins I have loaded?" and it will grep your sessions and check for you. Just axed like half my setup.

I've been messing with text to speech models as well and it's still kind of mindblowing that hosted ones are significantly faster than locals.

Been making good productivity strides with adding some new skills to my Claude Code setup, anybody have any good recommendations for what they use?

Really trying to use AI Coding tools more descriptively+spec oriented way this week and less of my super handholding and iterative way that I normally do. Using `/grill-me` skill to get a lot more of the domain knowledge and questions up front. Anyone else have any tips?

It doesn't matter if you use Claude, Codex or Grok. If you're not building in continual improvement from each loop through the code you're going to eventually drown yourself in bad code.

What's a better Text To Speech? Say, Kokoro or paying OpenAI for API calls? I'm hacking together my own voice assistant and built a comparison of 25 curated options that are all pretty good. Personally probably going with af_nicole for this project. voice-bakeoff.vercel.app/

I think this is still one of the things holding big corps back from really unlocking dev productivity. Every high output dev I know that's positive on AI is also doing dictation into their Claude Code sessions.

No, but that's a cool idea.

What are folks doing for in-repo documentation for AI building? I've bounced between Google's OKF, a docs subfolder, overloading AGENTS.MD Any Tips?

You can ignore the posts about Jev and whether LLM classifers are better if you remember one thing: always split your AI classification and your AI generation into separate tasks.

This is kind of like saying notepad.exe is the top programming editor. ubiquity =/= usage

Stupid low hanging fruit. Sped up our test suite by 5x by just asking Claude 'hey can you make the test suite faster'

I think I've just been burnt b/c every time I've worked with an org that's attempted to self host git (never mind their own internal CI/CD), it's had double digit downtimes that just completely killed velocity.

Claude, Codex, Grok, etc. are wildly inefficient running shell commands because they call the default verbose versions of them. Using RTK, I've been able to cut input tokens from them by _half_ - nevermind that this also speeds up everything as well. Recommended.