I built an MCP server for my voice assistant. By default, Home Assistant offers a few tools to the models, but I felt like it was missing some, so I built it.
It has tools to get real jokes, do math, unit conversion, get stock information, search the web with Kagi, and other things I consider to be essential in a voice assistant. It even got approved by my wife.
At work I use our MCP server to debug and troubleshoot issues with our code too. Just yesterday I used it to figure out why a tag wasn’t showing up in traces; using the MCP server, the LLM was able to see that the tag was present in parent traces, but was lost along the way, and figured out why.
Could I find that without the MCP server? Yes, but I’d take me hours of going through lengthy traces, while the MCP and LLM found the issue in minutes.
I got a Pebble Index 01 ring and it has support for custom MCP servers. I run one on Tailscale and it can interact with Home Assistant, my home issue tracker, my Anki setup, etc.
I've come full circle on MCP. With agents with a shell tool they quickly became unnecessary but these days I've been coming up with more agentic use cases where I don't want to give the agent a shell, and a narrower MCP interface is a great fit.
"What amps and charging schedule should I set my EV to charge at, so that at peak electricity rate time my house battery is at 90%." Looks at my house battery, car battery, current weather forecast, and house solar production, all of which are integrated into my Home Assistant. Looks at my gmail calendar. "16 amps, 9:30am-3:30pm, higher than usual because you're driving to lunch from 11am-12pm" (shows math). It can even go ahead and set it for me but I haven't given it read/write access to my car yet.
I don't know about others, but I use JSON because it's in the standard Go library, and most of the times, I rather use something weird than add a dependency.
I paid an actual designer to come up with a design system for my project, which I then made Claude follow. Works like a charm.
To give more details, the designer created the system in Figma and created a guideline document for a human to follow on how to use the system. I then gave those files to Claude Fable using Claude Design to turn it into code.
Then when I need to do anything design-related in the project, I make Claude follow the system. And I can use the guidelines to see if it’s following everything correctly or not.
With SourceHut banning projects that use LLMs[1], I’m looking to migrate this weekend.
While I can host Forgejo just fine[2], doing so means no one will be able to contribute to the projects, since federation isn’t a thing yet. Heck, getting contributions on SourceHut was hard enough, and it doesn’t even require people to register an account.
I might need to bite the bullet and go with GitHub. Regardless of what I think about the company and Microsoft, it’s where the people are.
Oh my. I hadn't noticed this, but I've been kind of expecting it. I've been a big SourceHut user for years now. So I'll be having to decide what to do myself. I absolutely get their stance, but I don't see a world in which I will go back to writing meaningful amounts of code by hand again.
I may try out tangled.sh. It's really nice to have my code stored and backed up somewhere other than infra I control and manage.
I'm personally pretty excited about Tangled (and ATProto in general), but I do have to caution that I've had some performance issues and some technical ones with hosting my own 'knot' and 'spindle' services
It ridiculous, Codeberg did the same. I get why, really, I do, but it's creating a severe damage that will be abused by a third evil actor that will take over the wasteland of GitHub and GitLab by creating a service that will allow anything (until it is enshittified of course)
Outcome? Everyone on their own self-hosted git servers, zero collaboration.
> Outcome? Everyone on their own self-hosted git servers, zero collaboration.
Everyone sticking to Github, then a year or two from now Github resolves their scaling issues and the competitors miss their window of opportunity to disrupt
Not if they keep putting everything behind a paywall. I'm not going to pay 60 (20/user/month) euros per month to be able to restrict pushes to main. It's absurd. Sorry, this specifically gets my blood boiling so I might be a bit more aggressive than needed.
What kind of collaboration were you getting on your random GitHub uploads? I didn't get any. This collaboration thing seems like an excuse if people have a way to contact you.
Codeberg only banned projects that are majority LLM-coded. They did not ban all use of LLMs.
It seems my eyes skipped the part of "[...]mostly consist of code written by generative AIs[...]" my bad. Which means I have to apologize twice, because of the other comment. Damn, good slap in the face xD
Not much to say, really; it controls my house in Home Assistant using an LLM, and a STT/TTS server. It acts as the Home Assistant Voice Preview Edition, with a trigger word and whatnot.
It's basically like giving an Alexa, Siri, or Gemini a body. It moves, show "emotions", comments when someone enter the room, uses the camera to see who entered the room and comment on that, make comments when my 3D printer finishes, things like that.
I can also turn on lights, the TV, the AC, ask about the weather, and other basic assistant stuff, ofc.
> It moves, show "emotions", comments when someone enter the room, uses the camera to see who entered the room and comment on that, make comments when my 3D printer finishes, things like that.
So basically you're using it as an AI mother in law, nice.
There is an application for Reachy that makes it so you can connect to Home Assistant, but it's broken and seems abandoned, so I forked it, fixed it, and customized it. That gets you 80% of the way. The rest is based on automations built inside Home Assistant.
And since Reachy's software and SDK are pretty open, it's trivial for a coding agent to learn how to communicate with the robot to achieve your desired outcome.
Fable is the only one that follows my instructions correctly, which I find quite important. It takes STYLE.md and SPEC.md as law, and code just like I would code, with the same mistakes and all.
I just can't get Opus (Opus 5 is dumb as a rock, to be fair) or Sol to do that, so I exclusively use Fable for personal work. When I reach my weekly limit, usually on the last day close to the reset, I just go back to coding by hand ¯\_(ツ)_/¯
Heck, it even does security reviews and fixes, as long as I don't ask it to "attack" the codebase. I'm planning on using Kimi or GLM for that part.
While I daily use an iPhone now, my original Pixel Fold is still working and running GrapheneOS. Besides a bug caused by GrapheneOS where the outer screen wouldn’t work, it has been flawless, and I can’t really blame Google for that one.
Giving it up was tough, as I consume a lot of books and mangas. I’m just waiting for Apple to release their foldable before I ditch this iPhone.
I review and approve every single line written by Claude because I want it to code like me, in my style. Yes, even with the bugs and problems I would introduce if I wrote it myself. Part of that is because I break from “idiomatic Go” from time to time because something reads better to my eyes.
For example, and this is minor but it annoys me, I like to declare variables in var() blocks if they relate to each other instead of using various := statements, and despite very clear STYLE.md, CLAUDE.md, and a personal style skill that is loaded on every session, Fable still insists on using :=, so I have to correct it from time to time.
It seems like auto mode would change that workflow. Hopefully, there is a setting somewhere to go back to the previous version. Otherwise, it might be time to finally give Pi a chance.
I review and approve every single line written by Claude because I want it to code like me, in my style. Yes, even with the bugs and problems I would introduce if I wrote it myself. Part of that is because I break from “idiomatic Go” from time to time because something reads better to my eyes.
For example, and this is minor but it annoys me, I like to declare variables in var() blocks if they relate to each other instead of using various := statements, and despite very clear STYLE.md, CLAUDE.md, and a personal style skill that is loaded on every session, Fable still insists on using :=, so I have to correct it from time to time.
It seems like auto mode would change that workflow. Hopefully, there is a setting somewhere to go back to the previous version. Otherwise, it might be time to finally give Pi a chance.
It has tools to get real jokes, do math, unit conversion, get stock information, search the web with Kagi, and other things I consider to be essential in a voice assistant. It even got approved by my wife.
At work I use our MCP server to debug and troubleshoot issues with our code too. Just yesterday I used it to figure out why a tag wasn’t showing up in traces; using the MCP server, the LLM was able to see that the tag was present in parent traces, but was lost along the way, and figured out why.
Could I find that without the MCP server? Yes, but I’d take me hours of going through lengthy traces, while the MCP and LLM found the issue in minutes.
reply