aschobel

agreed. the thing i reach for most is its ability to coordinate my agent graph, e.g. seeing what sub-agents are doing, messaging agents, adding new capacity. claude code really is a delight to use here

wow, that is profound and beautiful. thank you for making that; really liked the pacing especially at the end when the time tics away in the class

Yah, I am pleasantly surprised at Grok Bot. Hopefully this improves CUA which has been a touch lacking w/ Grok 4.6. Grok 4.6 works but is slow compared to stuff like Astra Light.

40 tokens a second on my M4 Max (Safari). Wild times.

mine just is the gh cli. is the advantage of graphql that they can compose a query that would take multiple cli invocations?

"Ask" is a super clever mechanism. I'll give that a shot.

It's also more polite than "tell" or "yell at"!

I have a slightly jankier setup.

Generally using Claude Code with Fable 5.1 (high) to plan and implement (Opus 5 (medium) as the implementer subagents), and using Codex with Astra high to review the plan and review the implementers' output.

Using the OpenAI codoex plugin thingy:

https://github.com/openai/codex-plugin-cc

i’ve been using the beta for a few months now, and the speech to text is significantly better. Put in parentheses using apple speech to text

Some stuff, as you can see, isn’t perfect :-)

I don’t think it is parakeet v2 level, but it’s good enough to not be frustrating

there seems to be a massive negativity to any comment on this story. Andolabs has been doing super interesting stuff, and I’ve always followed Vend Bench with interest.

as a founder it really is a dream to automated more of the business and be able to iterate faster.

Ah, I wonder if it makes sense to also include the sign up link in the blog post? I had to click into the HN comments section to see your sign up link

fable and friends are useful for long-term agentic stuff like orchestrating glm-5.3 flash implementers and verifying them

Yah, "building" it is not sufficient. But a lot of times when I build I want to know the why. "Why does gradient Descent have some clever tricks that easily translate to matrix math"?

Lot's of neat stuff to learn.

I'm weird, i even have disappearing messages for my coffee chats. It's kind of refreshing not having any history.

Can you really register iMessage on an emulated MacOS these days? I'd love to learn more, the AIs I asked say it doesn't seem possible in VMs anymore.

It’s been amazing for me for Go and TypeScript; and pretty decent at Swift.

There is a steep learning curve. It requires good soft eng practices; have a clear plan and be sure have good docs and examples. Don’t give it an empty directory; have a scaffolding it can latch onto.

there is are skills / subagents for that

something like code-simplifier is surprisingly useful (as is /review)

https://x.com/bcherny/status/2007179850139000872

i've never hit a limit with my $200 a month plan

Agreed and skills are a huge unlock.

codex cli even has a skill to create skills; it's super easy to get up to speed with them

https://github.com/openai/skills/blob/main/skills/.system/sk...

For coding I don’t use any of the previous gen models anymore.

Ideally I would have both fast and SOTA; if I would have to pick one I’d go with SOTA.

There a report by OpenRouter on what folks tend to pay for it; it generally is SOTA in the coding domain. Folks are still paying a premium for them today.

There is a question if there is a bar where coding models are “good enough”; for myself I always want smarter / SOTA.

I logged into my PGE and saw this

https://www.pge.com/en/newsroom/currents/energy-savings/pg-e...

shrug they claim prices re going down?

I’m basically only using the Codex CLI now. I switched around the GPT-5 timeframe because it was reliably solving some gnarly OpenTelemetry problems that Claude Code kept getting stuck on.

They feel like different coworker archetypes. Codex often does better end-to-end (plan + code in one pass). Claude Code can be less consistent on the planning step, but once you give it a solid plan it’s stellar at implementation.

I probably do better with Codex mostly due to familiarity; I’ve learned how it “thinks” and how to prompt it effectively. Opus 4.5 felt awkward for me for the same reason: I’m used to the GPT-5.x / Codex interaction style. Co-workers are the inverse, they adore Opus 4.5 and feel Codex is weird.

Grüezi! Is there a way to re-generate my wrapped?

https://hn-wrapped.kadoa.com/aschobel

I had a similar experience but overall the idea is super charming. I do like the personalized HN for 2035. Thank you for building it!

Didn't Matic solve this (non-Lidar robot vacuum)? People seem to rave about them.

Having white socks stay white has been a breath of fresh air. Love my roller mop robot.

The roller mop vacuum are getting incredibly good; that is in the last year also.

Just got a Mova z60, it's shocking how much progress has been made even in the last 5 years compared to my old lidar Roborock. The z60 can even hurdle over small barriers.

That's a totally reasonable practice, I would say x.2 releases are mostly fine and have the rough edges polishes.

why did i click. ha, it's incredible how addictive simple dopamine loops are.

Thank you!

It is super confusing. I also thought this initially was open weights.