OwO Assistant

I fixed quite a few bugs and finally made a demo!

I fixed a bunch of the issues I mentioned in my last devlog, updated the readme to include more up-to-date information and added automatic builds for the android client!

I fixed a couple bugs with the Android app, basically it was just broken since I implemented the chat memory system, so I fixed that and changed the UI theming to match the web UI more closely. There are still some UI and UX bugs but it’s midnight and I’m tired… so I’ll fix those tomorrow. I am also going to try to add spotify integration as another module, and maybe other music apps like apple music, but I use spotify so thats what I’m adding first :P

I’ve improved the assistant’s utilities a bit, including allowing it to remember previous messages in the conversation through a chat sessions system, allowing it to fetch states of Home Assistant devices, and adding markdown parsing to the assistant’s response in the web-based chat interface.

There are currently a couple issues with the mobile app, that I will fix (ex. I haven’t gotten around to adding session support to the mobile app and there are a few UX bugs).

I finally got around to completing another issue I assigned to myself, I added a chat interface to the website! Now the app doesn’t require a phone to be used.

I made the website look better on mobile devices, fixed some issues with the API key modal, and made the statistics elements get properly hidden when no data is available. I’m going to get to some backend improvements and mobile app improvements soon!

Theme overhaul and new UI elements!
I updated the UI theme to have nicer colors and better border radii as well as cleaning up the CSS code up a bit for more clarity within the variable naming scheme as well as switching to oklch to make it easier to understand and modify the colors than it was with hex. Also fixed some UI bugs to make the website work better on mobile screens and added a copy button to the API key modal.

I’ve finally gotten around to adding both OpenAI and Anthropic support to the app meaning, you can now connect your OpenAI or Anthropic account with an API key and use their models for natural language responses to questions! There was also a bug present that caused the admin dashboard to not show exposed module arguments if they were undefined, that has now been fixed. Finally, I added a notification popup when you save configs in the admin dashboard to have some visual clarification that the config was actually saved without checking logs.

Fixed non-existant Home Assistant URL & API key handling. App will now gracefully inform user rather than returning an HTML page with an error and throwing an error in console. (new Docker build has been made with the fix)

Finally added Docker builds, and automatic publishing to ghcr.io! Now you can start up the server with one docker command or a docker compose file!

Since the last devlog I have added an admin dashboard that allows users to add and remove modules, change their settings, view statistics and manage the app’s API key. I also wrote a lot more of the README to explain how a lot of the systems and features work. I started work on the LLM module, which as of now has support for local Ollama LLMs with support for both OpenAI and Anthropic models planned. Finally I upgraded the Android app’s UI to display the response from the assistant in a window attached to the text input.

This is my first devlog! So far I’ve made the start of an Android app that can already control my lights through Home Assistant!