bernhard

@wbmnfktr I'd love to understand what you are suggesting but I don't. Could you please be more descriptive? What problem do you solve? What is the goal? What is the solution? How would it help me with my day to day work? What would be the expected outcome? What would be the benefit (compared to other approaches)?

And you get useful results? I installed it in cursor and it was impressive to see the browser pop up, but not really useful... I even added an /auto-login route to RockDevTools to make the mcp login as superuser by default, but still it was not able to fix such a latte exception issue.

RockPageBuilder is free: https://processwire.com/modules/rock-page-builder/ But yeah, it's for sure not as performant as a single body field... You can cache the output via Template- or MarkupCache (free) or using ProCache (not free, but great)

Thx @Peter Knight ! This is so crazy! I asked it to add an onboarding tour overnight. The first two page impressions led to exceptions, but after copying those error messages to cursor it worked: And it not only works it also looks great! There are some issues still, but it's amazing how fast you can try things out and get a real world experience and not just a pencil sketch. Is it somehow possible to give the AI access to a browser so that it can try to load the page and see and fix such exceptions on its own? Another thing I'd love to have is to make it PLAN upfront but then start building once that is done. It seems I have to always confirm the build step after the plan is done.

Oh actually I completely missed that! Would be interested in more details how you do that or any helpful links 🙂

Ok 8 days later I've been more productive than ever before. I tried to use Opus via API directly, but as far as I understood you lose some integrations with cursor when doing so. So I switched back to auto-mode and realised that cursor actually has two separate quotas: The API usage was basically eaten up in 2 or 3 days (using Opus4.6), but I've been developing a lot during the last days - everything on auto using cursor agents. And what should I say... It's impressive! I've released new features for our SaaS app in lightning speed and I'm still only at 30% of my quota and I think the 20$ plan should easily be enough for me. --- I've also been testing Codex on my Mac and in the cloud. The idea would be to have agents running over night while I'm sleeping. Though I have not achieved any reasonable success with that approach so far. I think you can setup environments in the containers that these agents spin up and I think once you have done so the AI would be much smarter in fixing bugs that it produces and thus delivering much better quality?! Any of you guys already achieved something like this?

Another learning: while it might seem to take long to wait for results of an agent... you can spin up multiple agents and let them work on two different tasks (like two new unrelated features)

Thx! Had him in mind, but forgot to mention him! --- Update from my side: Had AI develop several features for my startup this week. It was fast. And it was good. Real quality code. Or let's say at least faster and better than I would have done it 😄 This is insane. My workflow is currently: Tell cursor to inspect the project and create the rules and skills necessary for agents to do their work Tell it to have a frontend developer for frontend stuff, a backen dev for backend stuff Spin up (multiple) agents and tell them what to do If it's a complex task switch cursor to "PLAN" mode first and check the plan before building Check if everything works Check the git diff If necessary ask for changes commit My learning so far: This is impressive - so far I have always had the opinion that the bigger the task gets the more AI struggles and it's better to do it on your own. I have always been a huge fan of cursor tab, which auto-suggests the next word or 2-3 lines of code, but this is another level! Good results cost money: I've been on my cursor 20$/month plan for a year and thought I was using it heavily... But now I've used 70% of my Opus4.6 quota in only a few days. The cursor pricing page states this at the moment: --> so I'd probably be ok with the 60$ plan or maybe even need the 200$ plan... Are you all spending this amount? I asked perplexity and it seems I can use Anthropic API directly in cursor and it might get cheaper? Any experiences/numbers to share @elabx or others?

Done. https://context7.com/phlppschrr/processwire-api-docs This is a quote from the other thread but I think it fits better here. @interrobang this looks impressive. Would you mind sharing more info about how that was built, how it can be used and how we can make sure we don't get prompt-injected something in chinese ^^

Hey @gebeer @interrobang @Peter Knight love your input but I think we are getting a little off-topic? I have created a new AI+PW thread here: Hope that makes sense!

Hey everyone, I've noticed that AI-related discussions are popping up more and more across the forum, but they're scattered across different threads and often go off-topic (guilty as charged). So I thought it's time we create a dedicated place to collect our experiences, tools, and workflows around using AI with ProcessWire. Why this thread? There are several existing discussions that touch on the topic: My recent post about Cursor turned into a broader AI conversation that drifted off-topic (link) There's a thread about MCP (Model Context Protocol) and ProcessWire (link) @gebeer started a thread about creating better Markdown documentation for ProcessWire - IMHO it was more of a request rather than a howto (link) All of these are related, but none of them serve as a central hub for the bigger question: How do we best leverage AI in our day-to-day ProcessWire development? What I'd love to collect here: What's your current setup? Which AI tools are you using (Cursor, GitHub Copilot, Claude, ChatGPT, something else)? How did you integrate them into your workflow? What works well? Where does AI genuinely save you time with ProcessWire? Module development, migrations, frontend templating, debugging, writing selectors, documentation...? What doesn't work (yet)? Where do the current AI tools fall short when it comes to PW specifically? Is it the lack of training data, the API structure, something else? Context & documentation: How do you feed ProcessWire knowledge to your AI? Custom rules, project documentation, Markdown exports of the PW docs, MCP servers? Tips & tricks: Any prompts, configurations, or workflows that made a real difference for you? Looking forward to your input!

Which parts does RockMigrations not cover? Not saying it's perfect, but from what you wrote I don't see anything it can't do.

It's a bit over a year since I started this post and it's crazy how far AI and cursor got! Huge shoutout to @gebeer who is my main source of wisdom when it comes to AI related stuff 🙂 I wanted to share a video that I watched recently. It is quite lengthy, but I was looking for exactly that to make sure I don't miss any basics. For me it was eye opening, so I wanted to share it: My personal takeaways/learnings: I have thought for a long time that AI is great to read/understand/explain or help me find something in my codebase but for actual coding it's not very helpful most of the time (other than cursor tab, which is awesome!) You can run multiple agents in cursor at once, and you can run multiple models at once, which is crazy The better you setup your environment, the better the results (obviously...) AI can write all your RockPageBuilder blocks with RockMigrations code easily You can actually TALK to cursor in the chat, which is a lot quicker than typing. You can even TALK to it in german and it will translate it to english on the fly, which is crazy. This is just the beginning So the question for me was is it really worth the effort of setting up your environment so that AI does the work for you and then in the end you spend more time debugging than if you just went ahead and coded it from scratch? I'm not sure yet, and I asked that question today @gebeer. He said clearly yes, and my gut said I can imagine he is right, but I did not really experience it myself. Today that changed. I took some time to try a new workflow for a client request. Usually I would have added code by hand, as that's a lot quicker. Why? Because you either start prompting and get bad results or you have to invest a lot of time upfront. But I know how it feels to invest some time upfront and then have superpowers forever 🙂 So I gave it a try and after a quick emergency rescue session with @gebeer I got very good results in actually very little time. And once I started to adopt that workflow it started to make sense more and more and I started to answer the question to myself: Is the effort worth it? Clear answer: YES So I want to encourage everyone to watch the video, try it out, ask for help if you hit any roadblocks. So why do I think it's worth it? Simple example: I asked AI to implement a new RockPageBuilder block. It did the backend for me quickly and easily. Then it did the frontend for me. The frontend needed more time and debugging, but I then added the uikit docs to my project and told cursor to add the docs to my cursor rules/skills/whatever (still confused, but cursor knows what to do). Then results seemed to get even better and it even fixed an issue in a way that I would have never thought of (because I didn't know this option was there). So I even learned something new. Finally I saw an issue in my .latte file, which is quite a common mistake: AI forgot to add |noescape to the {$page->headline} output, which made the headline "Foo & Bar" show up as "Foo & Bar". I guess every latte user knows that problem 🙂 The solution: Tell AI to fix the issue and to also add that info to the frontend-dev skill/rules. That means it will likely never make this mistake again! This is a really powerful approach. I'm quite impressed. And I think I'll need a more expensive subscription soon ^^ Which is one of the downsides. You have been warned! 😄😉 PS: The initial setup is also quite easy. All I did is to talk to cursor: A bit messy instructions and lots of mistakes, but it doesn't matter as long as cursor understands you. I'm quite sure the results are not perfect, but from what I read in the instructions for the AI it's a quite good overview of the project! And it's something to build upon and grow 🙂

That's not the case. If you release it as open source you are still the owner of the repo. Others can then "fork" the repo (which creates a copy of your repo in their own github account) and can then work on it, make changes and commit them (to their repo). Once they have new commits they can create a pull request (it's easy, the github website will suggest it and it's really just a click of a button). That pull request will then be added to YOUR repo and you can then merge it (or anyone having rights for it). @gebeer how is that approach different to what we have been talking today? is one approach better than the other? seems like it's doing the same thing?

That's because there is none. Or there are three - depending on the point of view. Some classes come from jQuery UI, some come from UIkit, and to make it even more flexible the new default admin theme comes with its own classes, conventions and technology (CSS variables instead of additionally to LESS). There is a readme for AdminThemeUikit here. If you are using the new default theme by konkat many of these instructions are obsolete. In that case I guess your best bet is to look for information they provided spread across the forum and the blog. But to be honest from what I read this is all just fancy marketing blabla with screenshots how nice the new theme looks, but nothing answers the questions you just raised. I basically asked the same questions here and got no answer. Maybe you have more luck! Oh, and if you expect that you can use regular UIkit markup in AdminThemeUikit you might be interested in this thread as well.