Bear Wang

Easy to adopt new technology but hard to adopt new ways of thinking.

Whether you use it or not is up to you, but it must be there.

@pierceboggan @code When can custom models be supported?

OpenAI models are now 12% faster in @code with our recent move to WebSocket mode on the Responses API! #_websockets-supported-for-openai-models" target="_blank" rel="nofollow noopener">code.visualstudio.com/updates/v1_118…

@zeddotdev This news is so sudden, I was wondering when the official version would come out a few days ago.

We've shipped more than a thousand versions of Zed, but all of them began with zero. Today, that changes. zed.dev/blog/zed-1-0

Introducing Hermes Curator! The new system built in to Hermes Agent now helps you keep your skills that the self improvement loop creates in check, by consolidating and pruning automatically. The curator does multiple things: - keeps track of how often you use each skill, when it was last updated/created, etc - Once a week runs automatically (configurable) - Uses the analytics plus it's own scanning of your skills and consolidates or prunes them if necessary - Skips externally installed skills, built in skills, and skills you "pin" that you dont' want touched. It will only attempt curation over agent created/updated skills or user written skills. - It will then determine whether skills can be consolidated, pruned, or otherwise made more manageable. It will convert some skills that are too specific into references, templates or scripts for larger/broader skills, or integrate them directly into a consolidation of an existing skill. You can also disable it entirely in the config.yaml and/or run it manually with `hermes curator run ` Learn more on the docs here: hermes-agent.nousresearch.com/docs/user-guid…

Doing the right thing at the right time doesn't always make you successful. Success requires a little bit of luck.

This task has been running for over 3 hours, what a terrifying context length.

The past couple of days, whenever I asked the Agent to generate a Plan for me, it often came out to over two thousand lines of English, taking a very long time to read. Going forward, I plan to have these Plans output in Chinese instead. Reading in Chinese is far more efficient than reading English—this has nothing to do with whether you're a native speaker or not.

@Dimillian This idea is very interesting, I am building a similar product.

A social network to share prompts and threads and comment them.

On Telegram, anyone can utilize AI bots to easily develop, launch and manage their own bot – with no coding required. #TelegramTips More information for developers is available here: #managed-bots" target="_blank" rel="nofollow noopener">core.telegram.org/bots/features#…

It'd be awesome if future AI could read minds—pretty bold assumption. Imagine AI could pick up on some thoughts and ideas in your brain using smart wearable devices.

openai.com/index/introduc… I really love the web design of OpenAI's introduction to Images 2.0, where the entire page is divided into two display modes: Image mode and Classic mode, allowing users to freely switch between them. This design is truly excellent!

After using Fish for nearly four years, I switched to Zsh yesterday. I really liked how Fish works out of the box, and its syntax is way better than Bash. But because it’s not very compatible with Bash, it’s not so friendly to agents. A lot of times, you have to remind the agent that you’re using Fish by default, which adds a lot of hassle in daily use. Our workflow has been changed so much by AI.

Today I spent several hours playing with Google Stitch, and I found that if you already have a reference website with many screenshots for it to refer to, its output becomes much more reliable. If you don’t have similar references, it becomes very laborious. The design I want to create currently has no reference materials, which is truly agonizing. I might give up and try using Codex’s latest image model instead—it might be a simpler approach.

After extensively using OpenClaw and Hermes Agent, I found that the best agent is actually Codex with the memories feature.

Next week I'm planning to systematically learn the basics of UI/UX design. In the future, software itself won't be worth much—what will matter is the design philosophy behind it, and how to make users love your product. Let's start from help.figma.com/hc/en-us.

@jesse_vermeulen Thinking about the next instructions

honest question: what do people do during the 5-10 min while Claude is running?

@alice_und_bob We are using the same keyboard.

Pivoting to posting feet pics now (Cat feet)

After learning about the story behind the TG team, I started paying close attention to their new features—each one's so well-designed, and they're rolling out updates super fast.

It would be more meaningful to show how much time you spent on product research and user experience instead of just bragging about how many tokens you used.

@imcryptohustler @Polkadot @hyperbridge Thank you so much for having such faith in the services Darwinia has built up over the years.

Another reason the EVM has vulnerabilities is that, unless other chains use Substrate or a framework that emphasizes safety measures, we should never trust third-party code. Instead, it may be better to trust experienced builders who have been developing since the Polkadot litepaper, such as @DarwiniaNetwork, @ringecosystem, and @HelixboxLabs. None of their bridges have ever failed—I’ve been using them for over 7 years.

We’re aware of an issue affecting @hyperbridge's Ethereum gateway contract. The exploit only affects DOT on Ethereum that is bridged through Hyperbridge and does not affect DOT in the Polkadot ecosystem, or DOT bridged through other bridges. Polkadot, its parachains, and native DOT remain secure and unaffected. Hyperbridge has been paused while the issue is investigated.