What people recommend, in their own words, with the receipt.
Austrian software developer, founder of PSPDFKit and creator of the OpenClaw agent; writes about AI coding agents at steipete.me.
21 things
AstraSoftware
Been using Astra for the last few weeks and it’s so good and proactive. There’s a bunch of PRs on vitest, tsx or SwiftPM where Astra debugged OC and ended up finding and patching issues in upstream dependencies.
On the model he has been running his own agent work through for several weeks.
x.com ↗·2026-09-03
Claude OpusSoftware Anthropic
I use it for most of my computer automation tasks and ofc it powers Clawd🦞.
After several paragraphs praising Codex, he says he still loves Opus as a general purpose model and it powers his own agent.
steipete.me ↗·2025-12-28
GhosttySoftware Mitchell Hashimoto
VS Code’s terminal is too unstable, had plenty freezes when pasting in large amounts of text. Nothing beats Ghostty.
A post laying out his current AI development workflow and the tools in it.
steipete.me ↗·2025-08-25
Dell UltraSharp U4025QWHardware Dell
3840x1620 makes 4 Claude instances + Chrome all visible without moving windows.
Says he is still extremely happy with his Dell UltraSharp U4025QW; the sentence quoted follows the monitor's name, which is a link on the page.
steipete.me ↗·2025-08-25
GPT-5.2Software OpenAI
GPT 5.2 goes till end of August whereas Opus is stuck in mid-March - that’s about 5 months. Which is significant when you wanna use the latest available tools.
Explaining why GPT 5.2’s newer knowledge cutoff matters to him when working with current tooling.
steipete.me ↗·2025-12-28
GoSoftware Google
My go-to languages are TypeScript for web stuff, Go for CLIs and Swift if it needs to use macOS stuff or has UI. Go wasn’t something I gave even the slightest thought even a few months ago, but eventually I played around and found that agents are really great at writing it, and its simple type system makes linting fast.
On which language and ecosystem decisions matter most when agents write most of the code.
steipete.me ↗·2025-12-28
SwiftSoftware Apple
Swift’s build infra is good enough for most things these days. codex knows how to run iOS apps and how to deal with the Simulator. No special stuff or MCPs needed.
Telling Mac and iOS developers they barely need Xcode any more.
steipete.me ↗·2025-12-28
CodexSoftware OpenAI
Codex is more like the introverted engineer that chugs along and just gets stuff done. It reads much more files before starting work so even small prompts usually do exactly what I want.
Explaining what he prefers about Codex over Claude Code.
steipete.me ↗·2025-10-14
Claude CodeSoftware Anthropic
TL;DR: Ghostty + Claude Code + minimal tooling = maximum productivity.
The summary line of his workflow post, naming the stack he settled on.
steipete.me ↗·2025-08-25
CodexSoftware OpenAI
So pretty much everything runs on high. GPT 5.2 and codex are close enough that changing models makes no sense, so I just use that.
A post on how his day-to-day building changed once agents, not typing, became the bottleneck; this is the model he settles on.
steipete.me ↗·2025-12-28
Jump DesktopSoftware Phase Five Systems
I usually work on two Macs. My MacBook Pro on the big screen, and a Jump Desktop session to my Mac Studio on another screen.
Describing the two-machine setup he runs agents across.
steipete.me ↗·2025-12-28
Mac StudioHardware Apple
I usually work on two Macs. My MacBook Pro on the big screen, and a Jump Desktop session to my Mac Studio on another screen.
Describing the two-machine setup he runs agents across; the Studio takes the browser-automation work.
steipete.me ↗·2025-12-28
ccusageSoftware ryoppippi
I used some token counting tools like ccusage and it’s all somewhat imprecise, but even if it’s just 5x it’s a damn good deal.
Working out whether a model subscription beats paying per API call.
steipete.me ↗·2025-10-14
VercelService Vercel
Pick services that have CLIs: vercel, psql, gh, axiom.
Advising that agent-friendly services are the ones with a CLI, and naming the ones he uses.
steipete.me ↗·2025-08-25
AxiomService Axiom
Pick services that have CLIs: vercel, psql, gh, axiom.
Advising that agent-friendly services are the ones with a CLI, and naming the ones he uses.
steipete.me ↗·2025-08-25
GitHub CLISoftware GitHub
Pick services that have CLIs: vercel, psql, gh, axiom.
Advising that agent-friendly services are the ones with a CLI, and naming gh among them.
steipete.me ↗·2025-08-25
AmpSoftware Sourcegraph
codex gets me far better results than amp. I have to give them kudos tho for session sharing, they push some interesting ideas ahead.
Comparing agent harnesses; credits Amp for session sharing while preferring Codex.
steipete.me ↗·2025-10-14
CursorSoftware Anysphere
Cursor… it’s tab completion model is industry leading, if you still write code yourself. I use VS Code mostly, I do like them pushing things like browser automation and plan mode tho.
Rating the editors and harnesses he keeps around.
steipete.me ↗·2025-10-14
GLM 4.6Software Zhipu AI
I keep an eye on China’s open models, and it’s impressive how quickly they catch up. GLM 4.6 and Kimi K2.1 are strong contenders that slowly reach Sonnet 3.7 quality
On Chinese open-weight models; the same sentence goes on to say he does not recommend them as a daily driver.
steipete.me ↗·2025-10-14
Kimi K2.1Software Moonshot AI
I keep an eye on China’s open models, and it’s impressive how quickly they catch up. GLM 4.6 and Kimi K2.1 are strong contenders that slowly reach Sonnet 3.7 quality
On Chinese open-weight models; the same sentence goes on to say he does not recommend them as a daily driver.
steipete.me ↗·2025-10-14
GPT-5Software OpenAI
GPT-5 also is much more literal in how you have to prompt it - it’s a great model but not the best agent, you have to be more precise and specific.
Comparing other coding CLIs and models against his Claude Code setup.
steipete.me ↗·2025-08-25
GeminiSoftware Google
so using it less and less. GPT-5 for reviewing plans works even better than Gemini.
On why Gemini dropped out of his rotation despite being a strong model.
steipete.me ↗·2025-08-25
Claude CodeSoftware Anthropic
I used to love Claude Code, these days I can’t stand it anymore
A hand-written guide to working with coding agents; here he explains why he moved off Claude Code to Codex.
steipete.me ↗·2025-10-14
CursorSoftware Anysphere
Cursor/GPT-5 takes FOREVER and doesn’t share its thinking, so it’s hard to steer.
Comparing other coding CLIs and models against his Claude Code setup.
steipete.me ↗·2025-08-25
Every entry is a verbatim quote and a link to the public post it came from. Nothing is paraphrased. If it is not on record somewhere public, it is not here.
The verdict is what they actually said: loved and liked are explicit; recommended means they told others to get it; uses means they only say they use it; read is a book on their public shelf with no verdict; mixed and disliked are kept too.
A dashed affiliate link, sponsored or their own label means the post disclosed a relationship. It is shown, never hidden.
The small bar beside a date shows how dated a recommendation is for its kind of thing.
The bar shows how dated a recommendation is, for its kind of thing. Full and green is recent. It drains, and turns amber and then grey, as the post gets older.
What counts as old depends on the thing. A laptop praised four years ago has been replaced twice; a book loved twenty years ago is probably still loved. The bar is empty after about:
A striped bar means the post carries no date and the date shown is an estimate. No date at all, no bar.