sunaookami

This is placebo and untrue.

HTTP response timeout is unused when TCP-KeepAlive is enabled (which it is by default): https://searchfox.org/firefox-main/source/netwerk/protocol/h... (also see https://bugzilla.mozilla.org/show_bug.cgi?id=1024015) and Firefox only uses the value until the FIRST byte is received anyway: https://searchfox.org/firefox-main/source/netwerk/protocol/h...

>I do also suspect firefox does some hidden retries if it can before it considers the download to have failed

There is no such mechanism.

You still get notified of a login and multiple times for the deletion request. Opening a chat app once in a week is not something rare and by that time you already changed your phone number inside Telegram.

I only know the German version from RTL2 which had funny commentary dubbed over it with made-up names and stuff. Curious, did the dubs in other countries do the same? Were there languages with the original commentary translated?

>But it confused the destination, sending the message to a Slack chat titled "Exec-team" — comprising XMTP's executive team — instead of my personal group chat with my AI agents.

>The CFO agent got the channel wrong because the channels had the same name.

Sounds incredible brittle, this should be linked via some permanent group ID (dunno if Slack has that) and permissions to only post there.

>Decision models have the potential to have an even larger impact on the Real World than LLMs have to this point

They really haven't.

You would still get a notification inside Telegram that someone else logged in (plus it sends the login code to your session first before you can fallback to SMS) and you can log them out. The "attacker" can't delete your account or log out your sessions because he can't use certain features on a fresh login, there is a downtime. But yeah, he can read all your chats. Same should be true for any app that uses phone number login. That's why there is a password feature.

--system-prompt-file replaces the prompt, I use it myself and the docs state it, too.

Then why bother using OpenRouter and paying the extra fees?

Use --system-prompt-file, it will replace the whole system prompt (https://code.claude.com/docs/en/cli-reference ). You'll have to use a shell alias or function or something to always append this flag when calling Claude Code but you don't need any MitM shenanigans. Then use "/context all" to see what else is sent (here I would recommend MitM'ing since Claude Code won't show the exact tools and text), there are a lot of tools no one needs and they are bloating the context, you can deny these in the settings.json (there is also a list here: https://code.claude.com/docs/en/tools-reference ). Also set "disableClaudeAiConnectors" to false to remove even more bloat.

>People prefer YouTube repair and cooking videos

Not sure about the cooking one but repair videos are horrible. People with thick accent, the worst camera, shittiest lightning and bad angles where you can't see what you have to do.

It's designed so that ads can be more easily integrated and make them harder to spot.

As if European companies are any better. Have you tried getting customer service from any European ISP for example? And companies still treat their workers like shit, it's just you can put up more easily with it since you can't get fired immediately.

I can also recommend hosting your own Gitea/Forgejo instance, it has an automatic repository mirroring feature that fetches the repo regularly. I use it to mirror some "high-risk" repos.

Gacha games do this, e.g. Genshin Impact. When you turn off your internet while playing a lot of game logic fails (e.g. damage calculation) before it tries to reconnect.

It lacks a lot of features compared to the "real" uBlock Origin: https://github.com/uBlockOrigin/uBOL-home/wiki/Frequently-as...

Two more weeks bro!

>When cargo fails to build and creates a massive amount of compile errors, you're better off having this preprocess step.

Claude already greps and tails every output by itself, a sub-agent would do the same.

Yep, I've used Pi in the past (see my other comment https://news.ycombinator.com/item?id=49908276) but I don't like the direction it went (selling out, forgetting their principles/throwing them out). And since Anthropic wants you to use Claude Code with their sub I just switched to CC again. I've only used Pi for a few months though when GitHub Copilot gave you 300 requests for like 10$.

I never hit compaction, most of my sessions are 150-300k tokens long with the longest being around 700k. Using sub-agents means that they can't use that cache, have to re-read everything and now multiply this for every sub-agent you call and it just wastes money/tokens. I also don't like how intransparent sub-agents are, I can't follow what they are doing and I can't really steer them. Claude Code has the /agents view but it's clunky and awful to use.

GPT context window is way too low for me and my last experience with it (GPT 5.6 Sol) was so awful and I hit limits way too fast that I cancelled it (and at least got my money back).

I'm no longer using Pi since it got worse IMHO and Claude subs can only be used in Claude Code but I miss the /tree feature which is perfect for first letting the model read & cache the important bits of the codebase and then start your plan from there (as long as you stay in the Cache TTL). Claude Code has /rewind but it's not as good.

I'm only using the 20$ plans.

Hah, it's the complete opposite for me :D. In Claude Code I disabled all sub-agents stuff, disabled nearly all tools but Bash, Edit, Write and WebSearch and replaced WebFetch with my own tool that doesn't summarize anything because the results were always worse with sub-agents, they always lack the necessary context and weaker models summarize bad. I also replaced the system prompt with my own that cuts a LOT of tokens, agents don't need a 10k+ system prompt anymore.

gpt-6-luna is terrible. It leaks tool calls and markers in the output like crazy, there is definitely something wrong here. gpt-5.6-terra works fine. Also, gpt-6-luna was sneakily added to the 1 mio free tokens group instead of 10 mio. like gpt-5.6-luna: https://help.openai.com/en/articles/10306912-sharing-feedbac...

And the Cloudflare dashboard is soooo slooooow and logging in takes an eternity, using Passkeys often just doesn't work ("not tied to a specific account"???) and the loging in button takes forever to be clickable because of some bot checks running in the background.

They should also improve the bookmarks and history UI, it's unusable. They redesigned the history sidebar now and it takes a lot longer to show up...

Naive and outright dangerous advice, every release fixes important security issues.

We rented VPS as kids

GPT models really have no taste huh.

Morphe just patches the YouTube app no? Then you need to still use that abomination. NewPipe/PipePipe is fast, lightweight and I couldn't care less about any "features" other than watching a video.

>invited me to a discord server

At least kids back then knew how to host their own TeamSpeak server ;) Clicking yourself a Discrd "server" is the easiest thing you can do.

They can't even use a computer, they struggle with basic computer knowledge because they grow up with closed ecosystems and most of them only have smartphones, not a desktop PC. They are called "Digital Naives" (sic) for a reason.