In progress: MELCloud Home & Classic ![]()
EDIT: already in error ![]()
EDIT2: it’s better now, thanks @pierre-gilles ![]()
In progress: MELCloud Home & Classic ![]()
EDIT: already in error ![]()
EDIT2: it’s better now, thanks @pierre-gilles ![]()
@mutmut Claude Code had disconnected, I reconnected the account, and restarted the factory ![]()
Thanks for making a request!
I really thought I was jinxed with AI!
MELCloud Home & Classic
First version tested and I left my feedback in the comments, it’s not great so far.
@pierre-gilles as you have a Mitsubishi and the old Melcloud (classic) app, I think, could you test it on your side if you have a bit of time? Thanks ![]()
Hi @mutmut ![]()
I’d rather let you iterate on your plugin to meet your use case. That’s exactly the goal of the plugin factory: to let everyone create a plugin that suits them, and then optionally share it with the community. ![]()
In your case, I’m not even sure a plugin that handles both APIs is really desirable. I’d prioritize one plugin per API: it keeps the logic simple, the code easier to maintain, and the evolution smoother, even if it means having two separate plugins.
I would have liked you to test it to see if it works as well as the integration you had done in Gladys (something I can’t do anymore because I have Melcloud Home and not Melcloud Classic).
In any case, I already know that I will very likely manage all the functions with my Broadlink (or Node-RED) because Matter doesn’t manage enough features for now.
@pierre-gilles could you create an Enphase topic starting from here please? I’m currently testing the integration with the factory.
thanks
Hi @ArMour85, I’m not sure I understand what you mean? ![]()
I created a topic about integrating Enphase solar panels.
If you launch an integration between My Renault and Wallbox, I’ll be able to test them.
Quick question: what’s stopping you from requesting the integration yourself? ![]()
Is the current form too complex or too long to fill out?
I notice there are fewer and fewer requests, and I wonder if the process has become a barrier ![]()
For example, would a site with just a “Plugin Name” field and a “Create Plugin” button help? Would it encourage you to propose more integrations?
For now it’s mostly a lack of time ![]()
Not for me, I’m used to using GitHub/GitLab, but for non-devs, yes, it’s probably complicated. What I’m missing is the reasoning behind Claude’s changes and improvements to the requests. In any case, I haven’t found it.
Yes, that will probably simplify plugin creation, but the sticking point is when you hit the limit of the LLM and have to get your hands dirty with the plugin code.
This is currently my problem with the Enphase plugin. I don’t know what to tell the LLM to move forward, and I don’t know the JS language or the plugin architecture to do it myself.
Indeed, you can help finish the plugin with the person who can test it, but it’s quite time-consuming.
At the end of each of Claude’s responses, you have all the steps of his reasoning!
You need to speak to it in French and just tell it what isn’t working concretely, as if you were talking to a human developer.
Claude Opus 4.8 is extremely intelligent, and frankly, it’s on par with a human developer. If you give it the same context you would give to a human dev, it will correct it just as well (if not better, and certainly faster).
The idea behind the factory is really to never look at the produced code, only the result.
Okay, I’ll think about it. It’s true that it could help democratize this tool!
I’m sure a more user-friendly interface would be a plus ![]()