I am doing a thing at https://gallotails.com
"inerte" is also my gmail username
I downvote anyone replying "correlation is not causation" to almost any scientific news. If you think the half a dozen PhDs with decades of experience involved in the study don't know that, you just wanna sound smart for internet points.
Claude Code goes through some internal systems that other tools (Cline / Codex / and I think Cursor) do not. Also we have different models for each. I don't know in practice what happens, but I found that Codex compacts conversations way less often. It might as well be somehow less tokens are used/added, then raw context window size. Sorry if I implied we have more context than whatever others have :)
Codex has always been better at following agents.md and prompts more, but I would say in the last 3 months both Claude Code got worse (freestyling like we see here) and Codex got EVEN more strict.
80% of the time I ask Claude Code a question, it kinda assumes I am asking because I disagree with something it said, then acts on a supposition. I've resorted to append things like "THIS IS JUST A QUESTION. DO NOT EDIT CODE. DO NOT RUN COMMANDS". Which is ridiculous.
Codex, on the other hand, will follow something I said pages and pages ago, and because it has a much larger context window (at least with the setup I have here at work), it's just better at following orders.
With this project I am doing, because I want to be more strict (it's a new programming language), Codex has been the perfect tool. I am mostly using Claude Code when I don't care so much about the end result, or it's a very, very small or very, very new project.
Maybe 10 years ago I interviewed both for Netflix and Facebook.
At Netflix, I used a state machine library to handle the project they gave me. Got rejected because I didn't show I knew raw JavaScript enough.
At Facebook, I wrote a calendar dropdown from scratch. Got rejected because I should have used a library.
Interviews sometimes is just a lottery...
Yes I know both companies should have set the expectation, but you can set the expectation for EVERYTHING, otherwise you give candidates all the answers you're expecting. There's always going to be some chaos due to the huge number of variables.
Possible, yes. Easier? I tried to search for YouTube videos of people doing amazing things at blazing speed using Gas Town about a month ago, and couldn't find any. I for sure didn't want to spend hours reading and learning something that I don't know if it even works?
Does anyone have like, projects built using it? I couldn't find "look at the output" types of videos or articles or repos, only "look at the input" types of posts about it.
This project is an enhanced reader for Ycombinator Hacker News: https://news.ycombinator.com/.
The interface also allow to comment, post and interact with the original HN platform. Credentials are stored locally and are never sent to any server, you can check the source code here: https://github.com/GabrielePicco/hacker-news-rich.
For suggestions and features requests you can write me here: gabrielepicco.github.io