Jolestar

During the Chinese New Year, I went back to my hometown and set up an internal network using Tailscale with several devices that were originally at home. The high-end desktop runs quantitative models and compiles Rust, the Mac Studio has OpenClaw deployed, Lazy Cat acts as a soft router, SSH is connected, and network wake-up is all configured.
While on the high-speed train, I remotely assigned tasks to the lobster, encouraging it to make full use of these devices. But suddenly, it stopped responding, and network wake-up didn't work either.
I called back home to ask, and it turned out that the

After upgrading Claude Code, encountering intelligence decline issues with third-party models. The symptoms are: 1. Clearly no typo errors, but it prompts typo errors; two words are identical, and if unresolved, it will randomly rename them. 2. Over time, it forgets the tool used to update files and keeps trying to edit files with sed, but complex matching with sed can't handle it, so it can't be fixed. It seems there are no problems if you don't upgrade.

My wife brought home an electronic piano for the New Year. I asked AI to write a command-line tool to control it, and I wanted AI to play the piano for fun. Because it needs to connect via Bluetooth, I implemented it in Objective-C. Later, I found that pure command-line operation couldn't maintain the hardware connection, so I changed it to an RPC service that the CLI connects to and calls. I then discovered that it was struggling with panics caused by multi-threading and memory management, which made me realize that the RPC service was also implemented in Objective-C. It seems AI also doesn't

I used a virtual credit card for over a year, but it was shut down again. Does anyone in the group have reliable recommendations for a trustworthy virtual credit card provider?

Recently, after the codex update, the permission mechanism was adjusted, and it feels very difficult to use. In default sandbox mode, bash is isolated, and network access is disabled. It's simple to get information via gh, but submitting PRs is impossible. If you want the agent to access the network, you need to grant all permissions. Additionally, search is not enabled by default and requires startup parameters. I really don't understand what the development team was thinking?

Teaching my wife to use Claude Code over the weekend.
She is a product manager and has never used the command line. I started from cd / mkdir / pwd.
Finally, after opening Claude Code, she basically didn't need me anymore:
She set up git through Claude herself, and even installed Docker.
The composability of command-line tools is brought to the extreme in the Agent scenario.
But this kind of composability also naturally comes with challenges related to security and standardization.

I used Gemini Vibe Coding to create an AI composition application. The idea is to use AI to generate sheet music and then convert it into music. Currently, it can only be converted to MIDI, but in the future, if integrated with a better mixing service, the results would be even better.
This is mainly to test AI's understanding of the mapping from sheet music to music. My testing shows it works pretty well, but my musical skills are limited. If any fellow music enthusiasts want to try it out and give feedback, that would be great.
Here is the demo video:
This is the trial link; you need to log

x402 v2 release. This is not just a simple upgrade, but a progression that transforms x402 from "a single implementation" into "a set of evolvable standards + pluggable reference implementations." This makes x402 no longer just an SDK, but truly an internet-native paid interface language.
In the v1 era, the x402 model was very simple:
Server needs money → Client signs and pays → Server verifies → Settlement completes.
This model is easy to understand but very "linear."
Once you need more networks, more payment methods, or more complex settlement forms, you quickly find yourself on a path of fo

Let AI implement a quite complex feature. I didn't have time to continuously review, so I used the Vibe Coding approach to let it add e2e tests by itself, then had another AI review and iterate feedback. In the end, the report was perfect, all functionalities were implemented, and there were no errors in testing.
Feeling great, I submitted a PR for Github Copilot to review. The result was that it created two versions of the core logic at the lowest level—one version implemented the code but didn't import, while the actual effective version was another mocked version😅.
The next challenge for V

The two paths of AI: competition for cloud entry or native applications on the user side?
Recently saw Alibaba release the Tongyi Qianwen App, officially entering the C-end app battle. Sam Altman has also repeatedly expressed his obsession with "super apps," continuously integrating various social features into ChatGPT.
In the narrative of major tech companies, the future AI app seems destined to be a replica of the current internet super applications - becoming the sole entry point and consolidating multiple functions into one.
But will the future of AI applications really converge on this on

Is Cloudflare down? Many websites are down, and ChatGPT is also down because it can't open those challenges. Or is it my network that's having issues?

Let AI reconstruct a large module, and it happily dismantled several modules, creating a new version. Then I had it run integration tests, and it tested repeatedly many times, but in the middle, the model quota was limited, so I upgraded it to the $200 tier. In the end, it deleted all the new modules and told me that using the old version was still more cost-effective. 😆 I was so angry that I started cursing directly in the chat box. It started working again and said it felt that the users were very angry...