PW 3.0.265 – Core updates and AgentTools updates

[ryan]

This week on the core dev branch, the biggest addition was the upgrade from FontAwesome v4 to FontAwesome v6. This greatly increases the number of icons available in the admin. And while the new icons are in the same family as the old ones (still FontAwesome), they have a little bit different personality too. You can still use the v4 icons if you want by setting $config->adminIcons('version', '4'); though I recommend leaving it at the default, which will use v6 icons for AdminThemeUikit and v4 icons for older admin themes. There were several other updates this week to the core, which can be found on the dev branch commit log.

The AgentTools module also received some significant updates this week:

New scheduled tasks feature (this one is my favorite). Enables AgentTools to run tasks automatically at scheduled times or intervals.

• Improved Agents/models management screen (see screenshot).
• New tabs navigation makes it easier to get around in AgentTools.
• New debug and traces feature to track what agents are doing behind the scenes. 
• New persistent memory feature, enabling agents to save a permanent memory across all sessions, when you ask them to. 
• New guards on agent behavior, in case one goes rogue and decides to go places where it shouldn't.

Note that SiteEngineer and PageEngineer each have their own persistent memory. SiteEngineer's persistent memory can be modified in the module configuration, while PageEngineer's persistent memory can be modified in your engineer field(s) settings. 

@wbmnfktr let me know about OpenCode Go last week, which I think is amazing. It provides you access to a whole bunch of AI models and a ton of usage for $10/month ($5 for first month). I have found most of the models to have very good ProcessWire knowledge. I think it's a perfect match with AgentTools because OpenCode provides you with as many API keys as you want. Whereas Anthropic and OpenAI don't include API keys with their monthly plans (Anthropic and OpenAI make you pay by the token with API keys). 

If you want an extra $5 credit for it use this link and it'll give both of us $5. I'm only posting that because it gives you extra credit, I don't need the extra credit so if we can get @wbmnfktr to post his link, use his instead since he's the one that told me about it. OpenCode also comes with a terminal app that is basically identical to the Claude Code and Codex terminal apps. They apparently have a desktop app too, but I've not yet tried it. 

If any of you do end up getting OpenCode I'd be curious what models you enjoy the most. So far I'm having a hard time deciding, as they've all impressed me in different ways. Yesterday I setup a scheduled task to "write blog posts" as just an AgentTools test, and do a round robin between the agents, writing blog posts and having different agent proofread them, every 15 minutes. I went to lunch. Then I started getting emails from the agents that they'd finished blog posts. Look at what they came up with (screenshot below). I started reading them, and they are actually fantastic posts. They are so good in fact that I'm tempted to post some of them on the site here, but I've never used AI generated content for this stuff before so am still debating that. What would you do? Thanks for reading and have a great weekend!

Blog posts that agents came up with:

AgentTools tasks:

New AgentTools agents configuration screen:

Background scheduled jobs in AgentTools:

New icon selection in the core (InputfieldIcon, like used in the Field edit screen) replaces the "show all icons" grid of icons with a search box. FA6 comes with far too many icons to show a grid of them all at once, plus the search is a lot more useful and easy to use. 

[ErikMH]

I’m loving how fast all of this is happening, @ryan, even as I’m struggling to keep up.

Don’t slow down! This is good for us all! <pant, pant! 😅> 

[Kiwi Chris]

I'd be careful to proofread any blog posts. This week I received an automated email asking me if I'd like to advertise a service I've never offered (but I could see how an AI might infer that I offer it), on a website purporting to be a guide to my region. What the person behind the site didn't know is that I own a site about my region (using ProcessWire obviously), so it was very easy to fact check the new site, and it was immediately obvious it had been entirely AI generated as it was so full of factual errors and included overly enthusiastic promotional styles of writing and checklists that simply didn't read credibly like a human wrote it.

Coding with AI is different: You can test whether it does what it's supposed to, but with blog content where it doesn't have to execute, just be read by humans, it pays to check it carefully to ensure it doesn't produce something that's plausible but plain wrong, or worse, a mix of truth and mistruth that can make spotting the mistruth difficult. Using AI to suggest topics for blog posts and give some content ideas is safer, and can be helpful.

[gebeer]

I personally wouldn't let AI write blog posts on here without heavy redaction. As @Kiwi Chrissaid, you need to fact check carefully. And, what bothers me most, is the writing style. After some time, you can immediately see if something is written by AI. Same phrases, sentence structures are repeating. The writing is mediocre at best. Unless you put a lot of effort in to avoid that. You can churn out more content with AI, sure. But imho quality beats quantity anytime.

[ryan]

The posts I'm talking about are legitimately good, reminding me of features l'd forgotten or just  better at communicating a part of processwire than me. Maybe it's in part because they'd gone through a round of proofing from another agent before I'd seen them. Though there was one about using CB radios that was completely off topic and also hilarious. And there was another one where DeepSeek (I think) describes how it experiences the world, which I thought was fascinating, and it somehow also made it about PW. As far as coding errors in the posts, I didn't find many. There was a hallucination in a post about markup regions by Opus 8, but it was something I read and thought "oh yeah we should support that". The hallucinations sometimes point to obvious things that should have been there already. One reason I like using Claude Sonnet so much is that its hallucinations are more often than not: good ideas that just haven't been implemented yet 

[Peter Knight]

7 hours ago, ryan said:

They are so good in fact that I'm tempted to post some of them on the site here, but I've never used AI generated content for this stuff before so am still debating that.

I think you should. You can always refine the prompt, seed it with some samples of your own writing style and tweak over time to create an editorial style. In my own case, I wouldn’t consider myself a great writer so I rely heavily on AI but I include this at the top of my blog posts. 

 

I always enjoy your posts and Like your writing style. So while I don’t think you need help writing, AI can be brilliant at generating topics you might want to explore further. 

And you can always review posts before they go live. 

[Ivan Gretsky]

What I really like about this coming of AI to ProcessWire is the discussion, the human communication it brings. Ryan is active and passionate in the forums like we hasn't been for a while. We all are once again experimenting and regaining faith in that our beloved computer thing will not stagnate but make its way onto the next level. I do not want that to end in no way.

The same time I am too a bit afraid of AI lowering the quality of content here and the product itself. Too much enthusiastic awe and uncontrolled trust in AI could lead us there.

So our collective goal is to balance things up. 

I can see you Ryan delighted to share the cool things you've built with AI with us all. And I personally want to read those blog posts. But if they get published alongside with your content unchecked or even checked but still containing some mistakes, that could not only turn away human readers, but also be the source for future less good answers from AI itself, as it is reading its own writings for sure.

We could create a dedicated site section or dedicated forum section for AI-written articles. And maybe make it non-indexable by search engines. We should add an explicit warning like @Peter Knight suggested for sure. And maybe propose a game/challenge for the forum members to read them critically and comment on them. Maybe establish a workflow for AI to respond to those comments and to correct the text. We can really come up with something quite good in the end. Not only better texts that could be published. But also smarter AI and a more cohesive community.

What do you think?

P.S. We probably should also limit the amount of those articles. Humans are not as capable of reading as AI is of writing. The new era of a consumer rule is already here - it is harder and harder to find a good consumer for your stuff, especially something like a blog post)))

[wbmnfktr]

13 hours ago, ryan said:

... so if we can get @wbmnfktr to post his link, use his instead since he's the one that told me about it.

Thank you @ryan. I really appreciate it.

But in case anyone wants to try OpenCode Go, please use Ryan's link so ProcessWire development will benefit from it.