Dev Palwar

@notcodesid whatchu studying?

big week

created this hero section for my side project. how's it?

wrote the full guide — exports, barrels, how to structure them, the traps, and the CLAUDE.md setup. medium.com/p/b803e9d696a6…

with a barrel they read one file and know the entire public API of that module. shorter imports mean less noise in the context window. named exports mean no ambiguity. and a CLAUDE.md with your module map means the agent knows your structure from message one.

A thread on barrel exports, a 10 min sol, and how this JavaScript pattern makes AI agents smarter. 🧵

@AnxKhn I can understand the kids but grown ass people are falling for such things lmao

The JEEfication of YC is happening

@SidTheITGuy @ezSnippet This guy knows the real nature of the Indian audience

- be @ezSnippet - Ask viewers to comment "TOOL" on a YouTube video, not even realizing that YT doesn't have that function - Throw wild claims like 80k a month in 3 weeks with just one SAAS tool and no effort. If anything he's the biggest TOOL out there.

@notcodesid How to use them properly?

people who are asking codex or claude? they dont know how to use them they are just doing the basic work with it. they don’t know how to use the them properly.

@N_and_ni Gaali dena gunah hai kya koi?

I usually ignore this, but look at it Is this really the kind of language people use? Look at the words he’s using. Why aren’t such accounts banned?

@yash_v_s you forgot this

Finally an end to: 1. “Anthropic launched claude opus 6.9 and the results are crazy.” 2. DewPieDie is the God and has figured out everything. 3. 1000 years in the same firm, how is it possible. 4. If you can’t put 4728463 PRs in a day, ngmi. 5. Look guyssss I got paid 6969$$$

@CodingNoobie @nikitabier @allegrajacchia i really thank you for your post man

hey @nikitabier and @allegrajacchia . I posted this screenshot because those comments felt manufactured but they are all very genuine people, I talk to them almost daily. They have been my friends for a long time now, they have always been very supportive. I only posted this as harmless banter that’s all. They genuinely post useful content for everyone. They even posted such comments because they all have worked together before. Please get the larger context. Please reverse your harsh action. Please ban the people being racist in the comments.

I taught Claude to talk like a caveman to use 75% less tokens. normal claude: ~180 tokens for a web search task caveman claude: ~45 tokens for the same task "I executed the web search tool" = 8 tokens caveman version: "Tool work" = 2 tokens every single grunt swap saves 6-10 tokens. across a FULL task that's 50-100 tokens saved why does it work? caveman claude doesn't explain itself. it does its task first. gives the result. then stops. no "I'd be happy to help you with that." no "Let me search the web for you" no more unnecessary filler words "result. done. me stop." 50-75% burn reduction with usage limits getting tighter every week this might be the most practical hack out there right now

@abbashali01 Idk man the only thing that matters to me is whether I want the service that website is providing. If it's necessary then I can't do anything but approve those permissions lol

I will feel so much more comfortable approving permission if companies actually provided more context about why they need a certain permission. It might be because I am mostly pessimistic regarding company intentions, so it always bugs me when they ask access to do so many things. Specially why I see no relation to what I want the product to do and what the product is asking permission for (which many times are legit cause shits often get complicated under the hood)

The fomo to ship fast is real. even when I try not to use LLMs and write code by hand it feels slow

How can I even trust the ai when it's entire concept is that it runs on predictions

I hope whoever invented PNGs like this is rotting in hell

spent 3 days vibecoding this app and now gemini realized that its an "anti-pattern" lol. Don't trust LLMs blindly guys always take decisions on your own and use them only for generating codes in a guided environment