Not sure what to think about Claude's "forced" 1M context. I was always paranoid to reset context when Opus/Sonnet hit 50% of the 200k window. More context =more hallucinations + more tokens burned. Did they actually solve that? Or do they just want you to spend more tokens?
Been working on adding an AI agent to the CraftMyGame game engine. Soon you'll be able to make a multiplayer game in under 5 minutes, without touching any code. Personalize it for your brand and community, and play it live with friends.
anyone know an app to record a "split screen" video (ideall 2/3 , 1/3 , with yourrelf and your screen?
pizza leftovers are the best
Probably why China might get ahead.
Now an official way to manage all Google service via CLI so it's easier for your (openclaw?) agents! Make agents life easier
LLM has been trained on hundreds, thousands using cli tools, they instinctively know how to query them
Would be cool if the Claude Code team just open sourced it at this point — if the layer keeps shrinking and it's all JS you can decompile anyway 😄
Feels like a buy vs build:
- Anthropic/OpenAI's harness auto-improves as the model evolves. But if it doesn't fit what you're building, you can't control the loop.
- Pi-mono is the same minimal bet (used by OpenClaw), vendor-agnostic and you own the harness.
How much do you want to own the agent loop? Claude Code team rewrites their agent harness every month, not because it breaks, but because the model keeps getting smarter and needs less scaffolding and directions(1/3)
Agents are like engineers. They need good tools to perform better. we're heading to a world where we start writing tool for LLMS to use them. Next step? Let llm agents writing their own tools.
coming soon , AI companions you can bring into their own video games 👀
Who knew I'd use the Stripe CLI instead of their dashboard?
I asked Claude Code to create products and pricing — didn't log into Stripe.
LLMs know hundreds of thousands of CLI tools. They instinctively know how to use them.
I've been toying with UI that morphs entirely based on what the LLM decides to show, and it's surprisingly fun to build.
Instead of fixed layouts, the model picks the components, the structure, and what matters most in the moment.
Making UI that dynamically changes based on what the LLM thinks it should show is genuinely fun
The model picks the layout, the components, the flow — all in real time. Generative UI is something else
Hot take: if your service isn't AI agent friendly, people will start leaving it.
Unless your UI provides something significantly better. Or even better, can your UI morph according to my needs?
How many services you use today are agent-ready?
just discovered this while using Claude Code across 2 iTerm windows with 5-6 split panes each. For once, Chrome isn't the memory hog in the room
is this a bug? can iTerm2 even scale Claude Code? is this why RAM is sold out everywhere?
AGI is here. We just can't recognize it because it doesn't always make the right decisions.
Like a fresh grad that studied everything for 100 years. But still (sometimes) missing wisdom, critical thinking, and judgment.
That's on us to provide, but for how long?
Always a delight to see how Claude Code decides to show you an answer. The form factor morphs every time based on context.
I was undecided on a design, so it presented variations to pick from. Every other frontend has a fixed UI. This one adapts to you.
am i the only one noticing opus is inferior to 4.5? tried codex 5.3 and it yields a lot better results than opus?
would you do it that way?
Vibecoding is fun until it's time to vibe debug . But why bother? just throw out everything and restart from a clean state.
Do you "--dangerously-skip-permissions"?
i'm watching claude code on the toilet instead of scrolling reels and it's not even close
Meta spent billions trying to get your attention.
Anthropic did it with a coding tool , how come?
Accidentally typed a Pokemon name into Google and a whole game appeared right in the search results. I have to search every of them to catch them all 🤣 Love these contextualized Easter eggs
Tried Claude Code's "Orchestrate Teams" to let AI agents collaborating. Agents stepped on each other's files, tasks went to wrong roles, tests all failing.
"What do you want me to do? Cleaning up the team and doing it myself, will be faster"
Claude being honest with itself 🤣
we can see which one is developers favorite 😂
That's not the future. That's today.
If you haven't tried a coding agent yet, you're missing out.
Give it access to your logs, metrics, and source code. Ask it "why is this service slow?"
Watch it pull data, correlate events, and point you to the exact line of code. Even better? let it do automatically 🤯
Hot take: SRE/DevOps is becoming about putting coding agents in the right places.
Incident → agent investigates (logs/metrics/source code) → points at root cause → suggests fixes → writes post-mortem.
LLMs are imperfect today, but give it a few years (or months?), they'll do that easily.
I believe this is a new way to think about how we build applications.
Coding agents don't just help developers—they enable entirely new kinds of products.
- Traditional: Product decides flow → user follows
- Agent-native: User states what they want → agent figures it out
I just wrote about it 👇
