Lately, I’ve been able to build small WordPress plugins with the help of AI tools (like Google Antigravity), and it’s been great. Now I want to move a step further—something like a WhatsApp OTP login or another practical Flarum extension.
The challenging part is that most AI models don’t seem very Flarum-focused. If I manually paste documentation, the AI can work with it, but that’s not scalable. What I really want is to build a setup where the AI already “knows” Flarum core docs and existing extensions, so when I give it a task, it can refer to those docs/extensions and help generate a plugin.
When I asked about this, the answer I got was: RAG + Rules + Reasoning.
Has anyone here actually tried something like this?
This doesn’t have to be limited to Flarum— any insights are also welcome. I’m really enjoying the whole vibe coding concept, but I want to get better and more systematic about it.
How are you all approaching this? Any tools, workflows, or real-world experiments you can share?
