But in practice you are padding token counts of agents reading streams of TUIs instead of leveraging standard unix pipes that have been around from day 1.

TLDR - your agent wants a CLI anyway.

Disclaimer: still a cool project and thank you to the author for sharing.

The problem is TUIs. Or at least the assumption you're only supposed to use a TUI. There are plaintext ways to invoke things but the advertised Claude/Codex experience is a TUI. You can't pipe the contents because it's not next - it's instructions to draw to the terminal.

TUIs are the new podcast! With everyone asking you to invoke tools via a TUI, there was nobody left to write actual tools.

TLDR - make sure you support plain text too ;)

So this being from github.github.io implies it's published by the "github" account on github.

On my most frequently used machine/dev env this means -

e for vim

m for mise

n for pnpm

c for Claude

x for codex

I (tend to) complain about actions because I use them.

Open to someone telling me there is a perfect solution out there. But today my actions fixes were not actions related. Just maintenance.

I would expect older models make you feel this way.

* Agents not trying to do the impossible (or not being an "over eager people pleaser" as it has been described) has significantly improved over the past few months. No wonder the older models fail.

* "Garbage in, garbage out" - yes, exactly ;)

I've grown attached to the git diff view, so I use it for reviewing PRs mostly (especially larger ones as github UI has been struggling with them as of late).

The rest of my code is written in Vim or by Claude.

I created the original plan with a very specific ask - create an abstraction to remove some tight coupling. Small problem that had a big surface area. The planning/brainstorming was great and I like the plan we came up with.

I then tried to use a prompt like OP's to improve it (as I said, large surface area so I wanted to review it) - "Please review PLAN_DOC.md - is it a comprehensive plan for this project?". I'd run it -> get feedback -> give it back to Claude to improve the plan.

I (naively perhaps) expected this process to converge to a "perfect plan". At this point I think of it more like a probability tree where there's a chance of improving the plan, but a non-zero chance of getting off the rails. And once you go off the rails, you only veer further and further from the truth.

There are certainly problems where "throwing compute" at it and continuing to iterate with an LLM will work great. I would expect those to have firm success criteria. Providing definitions of quality would significantly improve the output here as well (or decrease the probability of going off the rails I suppose). Otherwise Claude will confuse quality like we see here.

Shout out OP for sharing their work and moving us forward.

I enjoyed the read. Felt like Russell just thought about this question for a while, and shared his thoughts. It was very practical and enjoyable.

Disclaimer: I recall some "wow, we don't talk like that anymore" moments. And I didn't enjoy the hyperbole of the cover quotes. But the content of the book debunks those.

Thank you for your comment.

I'm interested in whether or not others agree.

On the day this was published (2025-02-07) it closed at $529.29. Yesterday it closed at $238.37.

Source - https://finance.yahoo.com/quote/IT/history/

Victory lap submission?

One discussion - https://news.ycombinator.com/item?id=44549436

I will admit I have waned enthusiasm a on Figma over the past couple of years. I find the UI churn confusing. The new features, i.e. dev mode and variables, feel out of place. I find the plugin ecosystem cumbersome. Doing simple things has become complex. I'm putting out real "who moved my cheese?" energy here I know. I suppose I'm wondering if others feel the same.

I'll be curious what happens come winter time. Midtown becomes gridlock in the evenings. I do not expect that to change.

All that being said - probably my own biases skewing things. I will keep my eyes peeled!

I had a much less intrusive incident but equally "alarming" ;). The apartment I live in has hard wired smoke/carbon monoxide alarms (with backup battery). One night, ~2am 1 of them starts beeping that the battery is dead. I confirm the breaker didn't flip or anything...finally decided to complete disassemble it and get some sleep. 4 of these in the apartment. Head hits the pillow and a second one goes off! Take all 4 down and live dangerously for an evening.

In the morning I do my research - turns out these things just start doing that after 10 years. "Get a new one!" it peeps incessantly.

Shame me please - as I bought replacements. I'll save this post for myself in 10 years so at least I can be the one to say "I told you so".