Quincy Mitchell 🔭

@badlogicgames I never have this problem and my guess is you’ve got some extension installed that’s slowing down the main experience. Try with minimal extensions? I happily use vscode daily / hourly.

am i holding it wrong, or has VS Code auto-complete just given up being helpful? why does it take seconds to surface the args?

@cursor_ai Make me feel more comfortable with secrets and agents.

Each environment now has its own version history with rollback. An audit log captures every change. Egress and secrets can be scoped per environment. Secrets in one environment aren't accessible from any other.

Starting today, you can run cloud agents inside fully configured development environments. Set them up the same way you'd set up a laptop for an engineer: cloned repos, installed dependencies, and toolchain credentials.

@brianmmdev @cursor_ai 👍 I’m excited to see more mobile things coming up. I wonder how similar it is to a tool I made for myself last night. Getting wild how easy it is to make stuff for yourself

Testing @cursor_ai from my phone. Not gonna lie, this is pretty neat

@badlogicgames I got most of my dreams from above done in one night. Used the SDK. Responsive Webapp companion app that is 3 panels for wide window on desktop , slims down to the below phone size.

@quincymitch in time

IT IS HAPPENING!

@mitsuhiko This is one of the many reasons why I’ve liked Cline. Rules = always in the prompt. Workflow = manually triggered slash commands. I liked this way of working before skills and still mostly have them instead of skills.

Bring back prompt templates ;)

@trashpandaemoji Hello - I risked trying PA today but I didn’t get far. I tried selecting the OpenAI-codex provider. As soon as I clicked “OpenAI oauth” button then no browser tab opened. Was like the button was not doing anything Any tips to get it going?

@badlogicgames My dream: Kick off an agent on a remote server. Steer the agent while it's spinning. Notifications when it's done or has a question. Review code changes and give feedback on lines of code for it to spin further. Phone companion app that has same features

super cool, that's the direction pi is going in too. looking forward to what's next at Amp HQ

@MichaelStonis @GeoffreyHuntley I’ve had c# be consistently worse than typescript for AI understanding it and generating it. imports aren’t as easy to track / properly reference. TS is an obvious filename reference, but C# namespace reference isn’t that straightforward. Have you seen similar?

@GeoffreyHuntley I've had this conversation a few times recently. I love C#, but also does any of the new syntactic sugar even matter? Maybe for me, but if anything it isn't what the models were trained on and it just muddies the water.

it makes absolutely no sense to me why everyone is mainlining nodejs/typescript still tbh. typescript is good for humans, humans no longer the target to design for.

@thdxr @zeeg I use it but I don’t like it. It’s just what is currently reliable.

@zeeg people said this when we had 100K MAUs and at 1M and at 5M we're about to hit 10M and people will still say this at some point maybe it's worth trying to understand what's happening

TUIs are not good sorry yall a CLI is a utility, and situational. this should not be confused with stuffing a full interactive GUI into a low capability platform. "lets ignore all the great UI technology of the last 20 years and build some caveman shit"

@mattpocockuk I’d enjoy seeing a few evals ( in my mind unit tests ) to help define improvement overtime for the skills themselves. Obvious that these skills are a good initial step but I’m guessing that evals can push them further & catch regressions as models change.

Nearly 23K stars for a collection of markdown files I wrote I guess they must be pretty good I want to invest more time in this repo. So, folks who starred it, what can I do to make these skills more obvious to you? - A docs site for the skills? - Send them to plugin marketplaces? Help me help you github.com/mattpocock/ski…

@trashpandaemoji Wow, inspiring! Thanks for sharing.. even if you never release to public, it’s still useful concepts.

@Howaboua I’d love to see this use sem diff

Nearly there, a couple more things and I can finally share what I've been cooking. Pi-desktop app that combines best of all worlds. You can already: - code with it (duh) - view diffs and send agents comments - commit, push etc. - use Pi either in GUI or TUI from within the app (for the extension-heavy workloads) - terminal - view inbox from all running sessions - manage projects & sessions - install, manage & create skills (global & per project) - install & manage extensions (global & per project) Snappier than Codex app. More suited towards Agentic Software Engineering - yes, not vibecoding. Roadmap: - chat view with basic websearch, artifacts and handoff to code - claw functionality - I have an extension for this already so will be a breeze, and there are plenty of adapters available - automations - because it's just a cron job - cowork - I mean, it's just skills, right? Already have a framework for this on GH. Hoping to give you a Pi desktop app you'll actually want to use. Thank you @badlogicgames for providing the framework. And I suppose @mitsuhiko for not letting Mario go insane by himself. Let's cook.

@badlogicgames @DanielGri Have you guys seen this repo? Would love to have the same UI but the diff content to be replaced with: github.com/Ataraxy-Labs/s…

so thankful to @DanielGri for building Glimpse. my shitty little pi diff review tool works on macOS, Linux and Windows without a hitch. Love it! Super trivial to build GUI extensions for pi with it. github.com/hazat/glimpse github.com/badlogic/pi-di…

@nickbaumann_ When should we make a new thread ? Or are you saying do the monothread thing with everything

@dshaw 👍, I agree with that in spirit but in practice the slop can be so large that it’s just noise. I’d rather see the outcomes we’re trying to achieve in the traditional fashion and for AI to just make it clearer. Out of curiosity or learning I’d want to see how the sauce is made

@quincymitch These days we're all handing each our team members work that's been touched by AI. I've found that sharing both provenance and process aids in the shared understanding of the outcomes we're trying to achieve.

I think AI is making individuals faster but teams messier. What do you think? 2-min survey. DM me if you want a copy of the results. 🙏 forms.gle/ihBgbZbsxUwnTw…

@Palanikannan_M I’d love to see this integrated with pi.dev , and have a similar experience as /diff-review , github.com/badlogic/pi-di…

line diffs are 20 years old. they still can't tell the difference between "renamed a variable" and "rewrote the logic." your diff tool doesn't understand your code, it never did. we fixed that, introducing Semantic diffs!

@Baconbrix 🙏 link to evals please 🙏 I’m trying to learn how to write good ones

The new Expo Agent skills improve native UI usage (headers, toolbars, sheets) by +46% in our evals 🚀 Add them to Claude Code: /plugin install expo@claude-plugins-official Other agents: bunx skills add expo/skills

@badlogicgames I’m just happy that your happy. Would love to help grow Pi. I had a dream last night that I made a vscode native extension for Pi.

i'm honestly surprised by the reactions to the announcement. twitter: 300+, all positive, with the Tolkien name being the only point of contention (see blog for my thoughts on that) bsky: my austrian possy, all positive, despite a general anti-AI stance, specifically from a societal perspective, which i partially share orange site: we don't talk about garry's site emotionally taxing. thanks all of ya for being so nice!

@Swizec There’s a larger interaction philosophy forming in my mind that this is a small bullet point example. Thanks

Markdown design and plan docs from your AI do not count as tangible results. The value isn't in the doc, it's in the work you put into thinking through the problem. I don't wanna read the doc, tell me how writing it changed your understanding of the problem