I know it sounds a bit unconventional, but finding solid, updated DNN modules can be tough sometimes. Lately, I’ve been experimenting with using AI and language models to help build a personal DNN module sandbox — a space where I can quickly prototype modules with some AI assistance.
For example, I recently created a simple WordPress plugin that pulls in the latest videos from selected YouTube channels and stores some of the data in a table for display: 🔗 https://azmanagedit.net/p...lugin-for-wordpress/
Now I’m trying to bring that same idea into DNN (9 or 10). Nothing fancy — just a lightweight module that fetches YouTube videos, stores key info, and displays them cleanly. It got me thinking: why not streamline the whole module development process using AI, especially for repetitive things like scaffolding and data models?
Has anyone else tried using AI tools or LLMs (like ChatGPT) to speed up custom DNN development?
Would love to hear your thoughts or see if anyone else is building a similar workflow. Quick question — does anyone remember which was the last version of DNN that included the User Online module by default?
At DNNConnect there were several sessions about leveraging AI to develop modules / themes.
https://www.youtube.com/watch?v=WN4DDnmEA1U was a great one from Sacha. If I find the time, I want this setup.
More specific on agents : https://www.youtube.com/watch?v=1lB1hVtJ1Oc from David.
I have had great success using https://www.cursor.com/ for module development. We have around 15 custom modules for our intranet. I find AI dramatically speeds up the process. However, as Will mentioned, it takes a little more thoughtful control of the context you send the LLM.
I will explain our situation in case it helps.
Our modules are generally small to medium SPA applications for internal business processes. Most are very similar designs, typically a front page with a table of data and then 3-5 screens to manipulate that data. Although we have done more complex modules involving image and document manipulation, etc. We use Will's https://github.com/Upendo.../generator-upendodnn to scaffold our project which includes multiple modules in one Visual Studio solution. We use VueJS for the front-end and C# for the API.
The initial few modules I mostly wrote by hand. However, I now almost exclusively use Cursor. I use AI to do the full stack development. I start by having the AI generate some very specific design documents based on my prompt and its questions. I spend a significant amount of time getting these documents accurate and concise because I make the AI refer to them for the rest of the development process. This and Cursor’s “rules” feature help establish guard rails for the rest of the process. I then work with it to generate the SQL scripts for the DB tables needed. Now I leverage my existing code. I basically tell it I want a “standard” poco/repo/controller for this DB object and because it has access to my other modules it can very quickly create the new API following my existing samples. For the front-end, 90% of the time it is the same thing. Follow patterns in my existing modules to create a view/form for this new object. I still code some. It still goes off the rails some. But it has gotten better quickly. Having standard solutions to give it as a pattern is a huge help. I use Cursor because it was the tool that had full agentic ability when I started. Carefully designing the “context” you feed it is vital to success. It still requires carefully verifying what it is doing. I have had it make changes that were completely unrelated and disastrous. I use my source control extensively and watch closely every file change.
These Forums are for the discussion of the open source CMS DNN platform and ecosystem.
For the benefit of the community and to protect the integrity of the ecosystem, please observe the following posting guidelines:
Awesome! Simply post in the forums using the link below and we'll get you started.