Onur Solmaz blog

follow: @[email protected]

Posts

I knew disappointment was around the corner, the flicker company being the flicker company

Such a small model, but so good at roleplaying (at least in this weird context)

Experimenting with SOUL.md on gemma4-26b-a4b (running on @DeepInfra)

don’t focus on the word “loop” so much, focus on “verifiability”

@maddada We've lost @thekitze now 😭

Slopus -> Fabulous

"Dogfooding caught the dogfooder" Fable 5 has a sense of humor

What did @karpathy see / was shown?

I swear to god, let this be a joke

To clarify, I was apparently on the enterprise team plan, which is I think equivalent to Pro...

Btw, there is no reason not to substitute this with a maxed out Macbook Pro Max as the...

What clankers are NOT

Keeping a workstation + terminal device is more feasible for running agents

Automations on Codex desktop app is really convenient for keeping track of @openclaw...

fun fact: Mario met @gvanrossum at 5 years old when Guido was hanging out at a cafe near his...

rwar

Not only I am further away from deciding, I am now considering Oppo Find N6 now as well, since...

I want to get a foldable phone to be my on-the-go control panel for all my agents

this makes be wanna vibe my own language

I am at @aiDotEngineer singapore, come and say hi if you are around!

i0 to i4 interest scale

This is also my setup now, except

And what good is this for? It lets me program my claw @dutifulbob to extract signal from the...

Keeping an INTERESTS.md file

First hand @OpenAI codex demos at @aiDotEngineer singapore workshops

Useful for automated constraints on your AI agent

Towards 1-click setup for local models in OpenClaw

This looks pretty cool!

It was a blast, thank you @clawcon @msg

Somebody *please* get this man a GPU

Emacsification of Software - Recommended read by @tqbf

Own your AI infrastructure

This is our moat 🤣

People's AI

> Worked for 10 hours

Gemma 4 first impressions on OpenClaw

gemma 4 is actually pretty decent and runs on my asus gx10 (128 gb vram)

@lucasmeijer @lucasmeijer one could actually periodically trigger an agent to propose...

Got tool calls to work, context size 65k tokens

@grok does this exist

Question for the community:

My clanker @dutifulbob has an identity update. I was getting sick of the despicable me theme...

OpenClaw type UK Drill Jazz Beat by @thorwebdev @GeminiApp @GoogleDeepMind

I will be there as well, speaking about ACP, acpx and agent orchestration 🙌

Full transcripts are available, including all the back and forth between codex, claude and judge

Repo link (feel free to send PRs):

Introducing AI Battle

Their argument “it’S HaRd On OuR iNfRa” so goes down the drain

AI replies are getting more sophisticated… or people are turning into AIs

The new github skill installed automatically by codex now causes it to prepend [codex] to each...

A throttling protocol for model providers

If your Claude subscription renewed too recently and you don't wanna waste those tokens, you...

If you dislike rotating ack emojis on your messages in openclaw, this is how to make sure it...

Plainer language for better agent UX

This has happened to some companies I worked at before

next up: claude agents sdk supports openai responses api 💀

Here is the spec and implementation for this flow. The mermaid diagram includes all the steps I...

acpx v0.4 ships Agentic Workflows

OpenAI early 2020s:

Wow even I as a frontend noob understand the significance of this

Token Leverage

Non-dev knowledge work needs version control

Another one, call me stupid: “How would Google have done it?”

MCP vs CLI

I just registered for an .agent domain and joined the .agent community!

Sep 2021 @lexfridman podcast with Don Knuth, they also talk about OpenAI Codex (code completion...

Damn I’m gonna have to switch to teams if it goes like that

Codex's long horizon task and instruction following has been the most life-changing AI feature...

Amazed everyday by the unreasonable effectiveness of in-context learning

This is unscientific, but there are certain keywords and phrases I use a lot while using...

Request for memes

Native support for Codex on OpenClaw

My takeaway from this is academia needs good social media and algo. For me, these...

Welcome ClaudeClaw to the Claw family!

Agent infrastructure needs Kubernetes

You should look into what Don Syme is doing at GitHub for automation with AI agents

Merge first, review later

@onusoz · 2026-03-20

Implementation plans keep agents on track

What I’m wondering after astral acquisition is, is OpenAI deploying Mojo internally, or...

Called it

Pro tip: tell AI to "explain in plain language" until you understand what you are reading

@onusoz · 2026-03-18

@onusoz · 2026-03-16

GitHub needs extensible repo rules for agents

Using coding agents in Telegram and Discord via ACP

Telegram/Discord is my IDE

@onusoz · 2026-03-07

AI security, the lethal trifecta, and Linus's Law

@onusoz · 2026-03-07

@onusoz · 2026-03-05

@onusoz · 2026-03-04

GitHub needs first-class support for working with agents

I built a coding agent two months before ChatGPT existed

The farming analogy for AI doesn't hold up

AI psychosis and AI hygiene

Anthropic's pricing is stupid

GitHub has to change

You don't need a skill registry (for your CLI tools)

Anthropic's pricing is stupid

Having a "tools" repo as a developer

Christmas of Agents

Depth on Demand

How to stop AI agents from littering your codebase with Markdown files

Agentic coding tools should give more control over message queueing

Vibecoding this blog

Google's Code Review Guidelines (GitHub Adaptation)

CLAUDE.md to AGENTS.md Migration Guide

Typed languages are better suited for vibecoding

Workaround for Claude Code running `python` instead of `uv`

Day 47 of Claude Code god mode

Predictions by Anthropic Researchers

SCP-3434: Istanbul Taxi Superorganism

Auto-generating pull request documentation with Claude Code and GitHub Actions

Working on the weekend

Don't delete to fix

Warmup and cooldown