Skip to main content

Activity for haglund.dev

Active on:

Loading activity...

Reply

i'm on the ultimate plan so my limits are unlimited searches and $270/year (the plan cost) worth of tokens so at my current usage i would end up at 60/270 used at the end of the year

Bluesky1d ago
Reply

I think the search limit in the cheap tier used to be a lot lower when I started using it As for tokens I've never done close to the limit before, and I'm pretty sure I've been using Kagi for longer than the usage details show

Bluesky1d ago
Reply

the model picker is also very nice :)

Bluesky1d ago
Reply

they count as separate things for usage i don't pay for any ai tooling specifically, i just use whatever is included in other subscriptions (kagi, jetbrains all products pack) but the assistant's ui and ability to search with kagi is incredibly useful for finding sources and verifying the answers

table of usage details from https://kagi.com/settings/consumption?range=1

Date (UTC) 	Searches 	AI Cost (USD)
Mar 2026 	399 	3.762
Feb 2026 	567 	6.743
Jan 2026 	848 	4.712
Dec 2025 	1,030 	5.868
Nov 2025 	544 	0.803
Oct 2025 	663 	0.477
Sep 2025 	643 	2.407
Bluesky1d ago
Reply

Try paying for a cheap plan first then upgrade if you feel it would be worth it to you :) Personally I burn through the cheap plans in like a couple days which is a good indication that paying for unlimited is worth it

Bluesky1d ago
Reply

Hope you feel better soon! Don't let perfect be the enemy of good/fun/improvement!

Bluesky10d ago
Reply

there is this option when right-clicking servers, but i don't think there's one per-channel

Notification Settings > Suppress @everyone and @here
Bluesky10d ago
Reply

I really wish the CompressionStream didn't use such horrific settings for zstd that it somehow does worse than gzip most of the time :(

Bluesky14d ago
Reply

this is what a quick script parsing the files and counting output the languages with 26 035 words are actually empty

console output of script counting words in each locale of monster hunter rise, for example:
english: 639323
japanese: 867301
Bluesky23d ago
Reply

you could always donate part of the profits to some charity if you want to help people :)

Bluesky1mo ago
Reply

The goal of a linter should be to offload mental load with autofixes and catch mistakes while teaching best practices. The limiting factor in my mind is time to execute rather than amount of rules :^)

Bluesky1mo ago
Reply

I wouldn't be surprised if I'm just misusing it tbh... 😖

Bluesky1mo ago
Reply

is it a choice to not render jsdoc `@example`s or is it just a library limitation? based on the generated docs for my little package `binary-util`

Bluesky1mo ago
Reply

Very nice! A big improvement from the current docs. Another idea that popped into my head was that hovering over something like this would also show what the strikethrough/whatever means, but that might be unnecessary since it probably also says that somewhere else below...

Bluesky2mo ago
Reply

I would probably some different combination of dimming and striking through them, both in the table of contents and titles of the sections

Bluesky2mo ago
Reply

Would it be a good idea to strikethrough legacy APIs instead of having a vague L badge next to them? Or maybe changing it from blue to yellow or orange?

Bluesky2mo ago
Reply

Nushell's `explore` is also pretty nice. :) `npm info --json vite | from json | explore`

Bluesky3mo ago