zilli

Hey, just a follow up here 🙂 Following the advice in this thread, I’ve been pretty happy with Codex CLI so far. I also tried Claude, but I ended up using Codex more. From what I’ve seen, both can do the job well. Most of the time, when something goes wrong, the problem is on my side 🫣 Usually weak planning, a vague prompt, or both. I’m still very much at the beginning, so there’s a lot to learn. But these are a few things that have been working for me: I treat AI like a very competent junior developer. Most of the time, I plan the task first and then ask him to execute. I might promote him to associate if things keep going well 🤣 I’m using the $20 Codex plan. You can actually get quite a lot done with it if you spend some time planning before asking it to code. I’ve just started using RTK to save tokens. I’m not sure yet whether it’s worth it, but I’m giving it a try. I’m comfortable with the CLI, so my setup is pretty simple: Arch Linux, Codex CLI, phmate, sqmate, and gog. I don't use MCP. Reading the model documentation and best practices has helped a lot too. It gives you a better understanding of how the model works. From there, it’s much easier to build your own workflow instead of just throwing prompts at it and hoping for the best. One of the most important things for me has been mindset, and I had to work on that quite a bit 😅 Before AI, building a new feature or even fixing bugs could take days, weeks, or months. Today, you can sometimes get something working in minutes. That’s incredibly powerful, but it can also be dangerous, especially for newer developers. The temptation to expect everything to be done in minutes or hours is huge, but reality is often more complicated. Maybe things will be different in the future, but for now, I’m perfectly happy spending a few days on something that would have taken me months before. That still feels like a big win. Finally, I wanted to ask @ryan and the others: are you using any additional guidelines, like the Karpathy Guidelines?

Thanks @ryan

I think this is the beauty of open source, right?

Hey folks, Quick question, is there anything you usually do to the database before pushing a project to a live server? Any kind of cleanup, optimization, that sort of thing? The reason I'm asking is that during development it's pretty common to install/uninstall modules, add and remove fields, rename things, test content... and after all that back and forth I'm a bit worried about leftovers sitting in the database that I might not even be aware of. Orphaned fields, stale module data, log entries, sessions, you name it. Do you have a checklist you follow? Any specific things you always make sure to clean up? Cheers.

A tiny ProcessWire admin utility. Hold a modifier key and click any Inputfield toggle to copy the field's machine name to the clipboard — without collapsing the field. Useful when building templates, writing selectors, or debugging: no more hovering to read the name in the title attribute. Github: https://github.com/dlzi/FieldNameCopier

Thanks for the write-up, @ryan.

Hey @ryan and fellow AI enthusiasts, This might be slightly off-topic, but I think it could be really valuable for the ProcessWire community if someone created a practical guide or tutorial on building projects with PW and AI tools. I’m new to this space, and honestly, it feels a bit overwhelming because there are so many different approaches and workflows. For example, after installing Codex or Claude, what’s the next step? Should I create specific .md files for context or instructions? If so, which ones? Is it better to have the AI read the entire ProcessWire codebase first, or should I simply ask it to build a module or website directly? I’d really appreciate any guidance, workflow examples, best practices, or general tips from people already using AI successfully with ProcessWire.

Hey @ryan, is this a side effect of your recent restructuring of the /wire/core/ classes?

@adrian noted. For me, the DB abstraction would open a lot of doors for us, special if we could use SQLite. I really want to make an app/desktop using PW and NativePHP.

Is a new module coming up? 🙂

@ryan It seems Claude is your real co-worker now—does that mean we can expect major developments sooner? I’m thinking of things like official nginx support, additional databases, organizing template files into subdirectories, and so on.

I'm testing it @adrian, and so far so good.

Hey guys — maybe you can shed some light here. I'm an old-school wannabe coder. My setup has been the same for years: Linux, PHP built-in server, and Sublime Text 4. It works flawlessly, and I'd prefer not to change it 🙂. Recently I've been experimenting with AI and I can clearly see how powerful and productive it can be. The thing is, I always use the web UI, regardless of the model. My workflow is usually: Use code2prompt.dev to package my code Upload it to the AI agent Start prompting This works reasonably well for me, but I often hit usage limits. From what I've seen, most people using code agents seem to have very different workflows (often involving Node.js tools, which I'm really not into 😞). So I'm curious: did you have to significantly change your development workflow to get good results from AI models? Or is anyone using something closer to a traditional setup like mine?

Thanks @monollonom

Hi all, I think I’m mixing up some concepts and would appreciate some clarification about when to use $sanitizer vs field text formatters. For example, I have a title field that already uses the text formatter “HTML Entity Encoder (htmlspecialchars)”. In that case, do I still need to escape it manually on output, like this? <title id="html-title" pw-replace><?= $sanitizer->entities($page->title) ?></title> Or should I assume that the formatter already makes it safe for HTML output, and avoid double-escaping? More generally: When is $sanitizer mainly intended to be used? On input, on output, or both?