michaellee8

Why cannot it just spend the inference doing the actual task lol

i found that at 700k-ish context even fable becomes an idiot, maybe openai's decision to cap codex context at 400k is correct, 400k is really a sweet spot where most part of the context is reliable.

I got the complete opposite of what you get, sol ultra literally vibed the entire system out for me from one plan mode approval.

is that called rust? that is the only thing i feel safe to let agents vibe code

if you have spend any amount of time in low level c vulnerabilities you will have heard about it, it is a very common time on the low level/cybersec space.

tbh i assumed that is an official product too

if you actually figure out enough pieces of bugs, even opus level model would be able to chain it together imo, and the latest china models has already been described as close to such level.

I guess some Tesla are manufactured in China lol. I am just trying to say that the liability that Chinese manufacturers takes aren't more than the US ones.

I think Google added that AI-generated responses maybe incorrect? When you are paying such a low amount of cost, like probably for free, I don't think you can expect a same level of quality as a human written or reviewed of answer. It is like same random user spin up their Lovable and vibe-coded a piece of slop and hold Lovable responsible for not giving them production quality code. It is simple, you get what you paid for. If someone actually figured out AI that is actually always correct, it would be charged in superhuman price as well.

That's not exactly the case in China, the current state of FSD is still pretty dumb, unless you consider transferring control back to the user at the very last minute before it crashes a proper way to handle risks.

TLDR:

SSH into a remote box:

go install github.com/michaellee8/notifytun/cmd/notifytun@v0.1.0

notifytun remote-setup # or ~/go/bin/notifytun remote-setup

On your own laptop/desktop:

go install github.com/michaellee8/notifytun/cmd/notifytun@v0.1.0

notifytun local --target [same-target-you-use-for-ssh]

Now you get Desktop notifications on Mac/Linux/Windows when your coding harness needs your attention. Same SSH connection you already using, auto-replay on reconnections, it just works.

---

I personally has an isolated VM to run Claude Codex/Codex on full auto so that I can leave it around and do something else, need a way to get notified when it is done, so I built this.

No port forwarding or sending your notifications to some random server, just the same SSH connection you already using, if you can SSH into the box, you can get notifications from it. Hooks setup is fully automatic. When you disconnects, notifications goes to a sqlite store on the remote box, so that it can be replayed when you reconnect (won't flood, just give you a summary if there are too many notifications).

doesn't claude code already store oversized output to disk and let the agent grep it?

I suppose they are vibe-targeting now

In that case I think you can have a refund subagent that is responsible for checking if the user really asked for refund before doing these dangerous things. But it only minimize errors, LLMs are non-determinitic by nature.

Just sent an connection invitation on Linkedin. This is actually designed for allow e2e automation using playwright-mcp for a previous startup i worked in that does voice-based job interview agents. The http endpoints is provided by a daemom sitting on the background, listening all input to the virtual mic and transcribing and storing it. The agent can hit /speak and /transcript through an mcp. We have built Livekit Agents specific solutions by injecting text responses but felt that is not enough since we want to be able to test the whole thing end to end so I hacked a way to do virtual mic/speaker. It was designed for closing the dev-test-debug loop so that Claude Code can develop on its own rather than relying on human to test it.

Interesting, I have built https://github.com/michaellee8/voice-agent-devkit-mcp exactly for this, launch a chromium instance with virtual devices powered by Pulsewire and then hook it up with tts and stt so that playwright can finally have mouth and ears. Any chance we can talk?

Probably not a good idea to let Claude vibe-selecting targets, it still sometime hallucinates

I only run software from Chinese companies inside a sandbox, either on my Android/iOS phone or inside a VM for desktop apps and only enable necessary permissions. Unfortunately Mainland tech giants have no sense of user privacy and would like to maximize their profit by collecting every single bit of your data because they don't profit on selling you the software, they profit on selling your data.

If they figured out it can be this useful in 2016 running 1 t/s, they would make it run at least 20 t/s by 2019

In that case we should have some sort of UI test backends I guess? This mcp was more for generic use cases which will allow any TUI framework in any language to work.

i think cs students should force themselves to learn the real thing and write the code themselves, at least for their assignments. i have seen that a lot of recent cs grads that has gpt in most of their cs life basically cannot write proper code, with or without ai.

the china government has been heavily subsidizing the electricity

fuck google for taking an ai-generated medium article as a reference and tell me that exceljs support charts (which it does not)

They do have lamda and it is available for test in their AI test kitchen. Seems much better handling of sensetive and offensive content then ChatGPT for me, but still cannot perform basic addition like ChatGPT does. I think it is technically better than ChatGPT but maybe they are only going to release the perfect product.

Tbf ChatGPT was far from production quality for serious applications, lots of misinformation and you can make it produce very offensive content. It is a good for toying around but you cannot take the output seriously.

Actually I am thinking of the EMDrive thing now. Would that be possible?

Is the HTTP API technically a wrapper of the vtgate's grpc interface?

That famous tech called cgroup was actually a Google contribution. But I agree that k8s is essentially Google's step to make themselves relavant in Cloud. They have missed the initial opportunity by promoting their PaaS AppEngine instead of something IaaS like ec2 in the beginning of the cloud competition, so Google just play the open-source game and keep releasing stuffs that can be used in all three clouds to lure people to use GCP. But then k8s is a very nice piece of tech that allows one to manage large clusters without vendor lock-in.

Web3Auth uses a WebAuthn + Social login approach to store the credentials. https://web3auth.io/

Disclaimer: I work there.

Actually Google did made a version of Google Drive Filestream for Linux, and made it available it Google Colab, just that they never released it publicly for unknown reason. I have grabbed the binary of it. https://github.com/michaellee8/gdrivefs/tree/master/assets

For example duckduckgo produce nothing useful when you search Chinese keywords. DDG maybe good at doing English searches but Google obviously have much more resources.