It is now more reliable, though it’ll still go down when hackai goes down and slack just stop responding every once in a while. It is now more clear in the testing guidelines, and I have improved the reliability of the region checking. I have also removed the feature where it checks for duplicates, as spam isn’t much of a problem (for now) and it makes testing hard.
I updated the readme with the required changes, and I also fixed the bug where it would say (undefined) if the printer didn’t specify filament stock.
After some feedback from some beta testers, i found that people want dimension filters, since most post companies only accept mail under a certain set of dimentions, and mail is much cheaper than parcel, especially internationally. I wired everything up to threejs and occt and the github api and got the bounding box. Now, there is also a /add-custom-filter command, allowing you to restrict it to any arbitrary set of filters.
I got it running on the server, but since most of my time was spent over ssh, that wasn’t captured by hackatime, but anyhow, it runs on the server now, and I also got a bunch of willing beta testers who are also printing for printing legion.
I got the database working, and the bot probably almost completely works. There is now a /add-as-printer command where you can specify the region where your receiving prints, and it will only include those. I also got eslint setup and refactored the whole codebase. Next, I need to make and upload all the assets like the icons and stuff and make good documentation. Also, I applied for a best server and I’m waiting for approval.
The bot is now able to parse print requests into a json schema using the hack club ai api. Currently the api is down, but it still kinda works with stealth models since they’re free, so right now, i’m using stealth/ox-alpha. I will switch to gpt oss 120b soon for speeeeeed. next, i’ll setup a json db (ik, eww, but it’s easier and im lazy) to store the requests.
I designed the bulk of the part for the paddle, but I still need to figure out what material to use and the method of manufacture, and if I choose CNC, I’ll need to do some more modifications.
Tools are too slow in high-latency environments, so, I’ve made it so that they output a block of code (that i’m calling a compute block) that in the future will be stored in a db that is unified for reference by future agents. I also refactored a bunch of stuff and got pasting stuff into the prompt box to actually work.
Next: I’ll implement the unified cache db. Or I might do something else. idk
I got tools to work, though I’m still figuring out where to get the best web search results, because exa and hacksearch are out of date and deprecated respectively.
I decided that the previous thing with a mobile app was a bit too hard and didn’t fit my application, so I switched over to a python textual tui. It now looks oddly similar to opencode, but the colours are the default from textual.
project is initialized with expo. I got the dev build running on my android. Now I can start making the actual app.
I made a readme and it is now able to run as an actual cli in .local/bin. I also figured out how to push stuff to github.
It should have the bare minimum done. It modulates sound and I made a selector for it (the gfw is blocking my requests to github, and my vpn stopped working, so gh only has part of it.)
