Bahit Hamid 🚀

@marcelpociot Aww yisss. Looks Polyscope would preset Linux with PHP development environment. I hope that is the direction. I miss Laravel Herd when I use Linux desktop.

Oh hello there

@marcelpociot Yay! Good job! We are now covering all OS. Will test this soon. 💃

Polyscope is now available on Windows 🔥 getpolyscope.com

@nikuscs @conductor_build I use @getpolyscope with Z.ai's GLM via their Coding-Helper tool so I can use it Claude Code. I stop using Anthropic's LLM last year and exclusively use GLM with Polyscope. I love Polyscope getting new features in every updates and it gets better over time. ❤️

Chat, who is solving AI Agents + Mobile review? Iv been trying to get away more and use the phone to prompt and review more but still cant get the perfect workflow ( yet ) - @conductor_build is amazing, but cant get mobile yet, also a bit hard to review changes - Codex is super nice + remodex but its more GPT only focused. - @superset_sh has been my daily driver, but no mobile app yet ( also priced at 20$ extra per month which is legit but another sub ) - Plannotator is amazing, buts its more local first/per session i guess? - Solo by @aarondfrancis is also nice, but no mobile/review as far as i know. - @getpolyscope looks nice and checks most of this boxes but worried about the Anthropic/Claude code part of it, would love it to be terminal only + The only candidate i saw that as potential that checks all of this is Passeo but its still very far from polished. Honestly iv been using screens 5 to get into my mac/mac mini and get the job done, but wondering who is solving this? Anything that i dont know about ( yet ? ) Please share for reach!

@paraddox What the.. I didn't realised they raised their price. I just renewed a few days ago and I was getting about $81/quarter year ($324/year) for Pro plan. Now it cost $194.40/quarterly. I guess I got the the old pricing. Phew..

Last year before X-mas I got the yearly Glm Max coding plan for 350$ Now it's 4x that. Ouch! and lucky me!

@basil_k15_ @simonhamp @getpolyscope getpolyscope.com/docs/getting-s… - I use this on Linux then use a browser to view it - but I think this is only available once you subscribe to Polyscope.

@simonhamp @getpolyscope why polyscope doesnt have a linux support?

that's it my babies, work! work hard and you shall be rewarded @getpolyscope

- Supervisord/systemd start with minimal env -> Node + helper invisible - no models Fix: run via a login shell with NVM loaded: `/bin/bash -lc 'source ~/.bashrc && polyscope-server'` (used above for command=) Lesson: tools relying on shell init fail in clean process managers.

Debugging Polyscope + Z.ai on Ubuntu 24.04 - Polyscope worked manually - Failed under supervisord/systemd Problem: models not loading Turns out: - Z.ai/Claude helper lives inside Node (installed via NVM) - Requires -/.nvm/nvm.sh to set PATH & env

Hmm I just tried downloading from a VPS in Singapore, it can downloading it fast with no issue. Perhaps it maybe just the ISP or just Brunei region in particular. Anyway, I'll try this Standalone Server soon. 😁

There is something wrong with Digital Ocean's France server and trying to download it to Brunei. I'm barely able to finish the download before timing out.

@marcelpociot @getpolyscope I am trying out the Linux standalone server of Polyscope. Tried to use the curl command for the script, somehow timing out. Then I tried to manually download the binary - man it is awfully slow. Even it is just about 100MB. Downloading it to Brunei.

@marcelpociot @getpolyscope Awww yisss.. Going to try this soon. Good jobs and keep em coming.

You can now setup @getpolyscope on your Linux VPS and use it as a headless server that is available 24/7! Connect to it via our website - either from your phone or from your desktop. getpolyscope.com/docs/getting-s…

@joshcirre Waaaaaaatt? You can do that? This would save me time explaining to the AI. Going to try this next. I wonder GLM would work reading screenshots via Claude Code. I haven't tried this. Thanks!

What if you could write all of your UI feedback across your entire app, take screenshots, and deliver that feedback to an agent without having to type up all of the context? Basically what if Agentation worked for any stack? I built Instruckt. Tailored for Laravel.

@marcelpociot @getpolyscope Was this close 🤏 to pay that $99/year but since I work in multios (MacOS at work, Windows and Linux at home), I'll wait until all major platform supported for this mobile AI orchestration. Keep it up. This looks promising.

Fixing bugs via @getpolyscope on my phone while sipping my morning coffee The future is good, friends

@marcelpociot @getpolyscope @laravelphp Just tried this morning, finding a few bugs: - if the "dashboard" is not connected (the green color at the status bar not green) any prompt written will not be sent and just get lost. - renaming workspace was not consistent (still had the random name as the Herd's url name)

When you add a new repository to @getpolyscope we automatically detect if the project is served by @laravelphp Herd and recommend a polyscope.json file. This way, creating new workspaces will always use the correct PHP version, and be immediately available in your browser 🚀

Trying out Agents Orchestration: - Polyscope by @marcelpociot ✅ - Superset.sh 🚧 - jean.build by @heyandras 🚧

@PovilasKorop I just tried @marcelpociot Polyscope today on a few repos and now I'm seeing this superset.sh - at a glance I can see documentation on superset needs more improvements, Polyscope has more focus in Laravel and Herd and tries to differentiate by using copy-on-write.

Another multi-agent software, they keep popping up! Superset - Run 10+ parallel coding agents (Claude Code, Codex, etc) on your machine superset.sh/?utm_source=pr… Seems to be free and open-source, with 5k+ stars on GitHub.

@JohnnyNel_ My guard rails that I think I could put my Laravel would be Sentry and using browser testing with Pest. Somehow the AI is struggling getting it done properly, I had to troubleshoot as it hallucinates the test commands.

@bahit vibe coding's been speeding things up for me... but knowing which guardrails to add after? that's the real game changer

Today I shipped my first vibe coded Laravel project - a Web app that capture PDF documents submitted by students during their internship program. Mainly powered by Z.ai's GLM via Claude Code (through z.ai's Coding Tool Helper for setup)

@DanielsBuilds Thanks. Its just me practicing how a work flow with AI, getting my feet wet and fixing a manual process that happens every semester at school. So far it's OK. Right now just struggling with FrankenPHP with Caddy and the best practices on using it.

@bahit Congrats on shipping your first vibe coded project! 🚀 That PDF capture system sounds super practical. The shift from planning to "just build it" with AI is incredible - you probably saved weeks of setup time.

It seems like I found gap in finding tutorials or steps on how to do this. Even GPT answers on this look like they are hallucinating. (I'm guessing the culprit here is the docs on both FrankenPHP or Caddy don't do much for a complete example for Laravel).

Struggling to use FrankenPHP with Laravel for production. Reset from scratch a VPS about 3 times. Challenge: use FrankenPHP without Docker with Laravel Reverb on the same server. Main effort is to understand what settings to choose for Caddyfile for Laravel production.