Pumped to join promptfoo as an engineer. I'll be working on AI agents/evals/red teaming/security for open source + enterprise.
Dane Schneider
@danenania.bsky.social
Developer and startup guy. Founder of plandex.ai, an open source AI coding agent CLI for real world projects. Also founded envkey.com, end-to-end encrypted secrets management. Now winding down after a 7 year run. YC W18 batch.
#buildingpublic two questions for you: - what are you working on right now? - from 1-10 how confident are you that it’s what you *should* be working on? i.e. that it’s the most important and impactful task on your todo list.
atproto question: this account is banned on the default bluesky server, and what does that mean exactly? if I hypothetically wanted to see their posts (not saying I do) they can still post to a different server and then I’d have to import or follow that other server somehow or…? How does it work?
Libs Of TikTok has been banned on Bluesky
#buildinpublic where do you live and how do you like it there?
Working with LLMs gives you the “works on my machine” issue times 1000. You either need an eval suite that covers everything or to manually test every prompt variation ~100 times. Otherwise it will definitely fail in all kinds of crazy ways in prod. #buildinpublic
can I code fast? no. but can I code well? also no. but does my code work? alas, no
Anyone working on LLM curated bluesky feeds? Is it possible with the api? #buildinpublic
Coding with LLMs highlights a key distinction that runs through all of programming: knowing what to do vs. knowing how to do something. LLMs are extremely useful when you know what to do, but don’t know how to do it. But if you don’t know you need something, very good chance LLM won’t mention it.
All the cool kids are doing it, so I made a starter pack, too. This one’s for the founders out there doing more with less. If you’re solo or a small team making big things, I’d love for you to leave a comment so I can add you! go.bsky.app/ChrSqgi
“Execute some commands for the user” sounds simple but there’s so much hidden complexity to making it smooth: - what OS? - what shell? - env vars? - cwd? - source profile files for global commands? - redirect stdin? - stream output or wait and show it all at the end? - error handling? - signals?
Spending a month on Bluesky has created a whole new model of a fight we need to be having right now. Not liberal vs. conservative or whatever, but respect vs manipulation. I am tired of being manipulated everywhere I go. That’s why I like it here, I feel like I’m in charge for once.
#golang is great for coding with LLMs. - Verbosity is much less of an issue - Compiled + types means codegen errors are caught quickly - Discourages abstraction and complex types, which LLMs struggle with - Culture of stable APIs, helps avoid version mismatches in generated code
Really enjoying Scavengers Reign—an animated scfi series on netflix about a planet teeming with alien life. It’s amazingly unique and creative with fantastic visuals. So many cool (and terrifying) life forms.
#buildinpublic devs: do you like CLI and terminal tools or do you prefer to do everything in an IDE/other GUIs?
Hey #buildinpublic how do you balance between iterating on small improvements vs. working on larger, more impactful features that take longer to ship?
I finally created my first starter pack for #buildinpublic #indiehacker and #founder who are building tools related to AI and LLM. If you are building AI stuff, I'd be happy to include you in 😁 💡 Share what you're building in the comment 🧡 Like and repost for visibility go.bsky.app/UcofkF4
Any starter packs for AI codegen I can follow or join? Interested in the full spectrum from research to prompt engineering, tools, etc
In my experience LLMs are not replacing coders but creating many _new_ coders who now feel much less of a barrier to jumping in. Thing is, once you generate that first app and start tweaking, you’ll end up all the way down the rabbit hole before long. Whether you want to or not 😅
Just signed up and I like the vibe here a lot, but one thing I think X has right is showing number of views on posts. It’s helpful when engagement is low to know if it’s just because barely anyone is seeing the post. Feels less discouraging that way. Any thoughts #buildinpublic folks?