moy yewhon

💙 Army

[Anatoli Kopadze @AnatoliKopadze]

an OpenAI engineer just showed how he gets agents to do his whole job: code, debug and more, using loops

29 minutes from the engineer who coined "harness engineering"

he writes the rules, agents write the code, a reviewer agent loops until it's right

the winners won't have the smartest model, they'll have the best loop around it

watch it, then read the full guide on loops below

[Raytar @Raytar]

My friend applied to 200 tech jobs in two years. No CS degree. No callbacks.

Last month Anthropic offered him $750,000.

All because of one Stanford lecture. Free on YouTube. One hour.

A professor explains how ChatGPT actually works. Not the Twitter version. The real one.

He watched it in bed. Paused it eleven times. After that hour he told me something I didn't believe. "It's embarrassingly simple."

Three days later he applied to Anthropic.

Every single question they asked him, he knew from that video.

[Raytar @Raytar]

Met a guy making $1.6 million a year.

Three days ago he was at a Meta conference. Told me he saw the best AI talk of his life.

Boris Cherny was on stage. Showed how the Anthropic team actually uses Claude day to day.

Boris deleted his IDE eight months ago. Now he codes from his phone.

I watched it last night. Had to pause it twice.

Not because it was hard. Because I realized I've been using Claude like a toy.

He sent me the recording. It was never published.

Posting it below.

[Rahul @sairahul1]

A senior Google engineer dropped a 424-page doc on agentic design patterns.

424 pages.

Most engineers bookmarked it and never opened it again.

I read the whole thing.

Here are the 15 patterns that actually matter — explained in plain English, with exactly when to use each one ↓

[Khairallah AL-Awady @eng_khairallah1]

Google CEO, Sundar Pichai:

"If you don't teach your agents to debug themselves now, you will keep wasting hours every week."

In 30 minutes he explains why the best engineers stopped writing code and started building agents.

Most people think building an agent requires an engineering degree.

It doesn't. It requires one guide and one afternoon.

Watch the interview, then save the exact setup below 👇

[Mr. Buzzoni @polydao]

THIS NOTEBOOKLM + OBSIDIAN WORKFLOW IS THE FASTEST WAY TO BUILD A REAL SECOND BRAIN

> most people dump sources into folders and never touch them again

the actual workflow: NotebookLM takes up to 25 million words of raw input - PDFs, YouTube videos, websites - and lets you interrogate them in a single chat with source citations for every answer

> once you understand the terrain, the best ideas move into Obsidian. the notebook gets archived.

NotebookLM is the scout. Obsidian is where your universe of ideas actually lives

📁 Obsidian
↳ https://t.co/0Hq2Bmh6jO

📁 NotebookLM
↳ https://t.co/YJYHE2kHL5

[Anatoli Kopadze @AnatoliKopadze]

Anthropic engineers just showed how they build a full app from scratch, using a loop of agents

40 minutes from the team behind Claude Code

they used three agents: one to plan, one to build, one to judge, cycling until the app actually works

the winners won't have the smartest model, they'll have the best loop

watch it, then read the full guide on how to actually use loops below

[Khairallah AL-Awady @eng_khairallah1]

this is f*cking gold

How to build your first AI agent (Full guide)

if I had this a year ago, I would've shipped my first app in a day instead of 2 weeks

in the right hands, this changes everything:

[Jamin Ball @jaminball]

Great read!

[Rahul @sairahul1]

🚨 Karpathy was right.

He warned that 90% of AI advice dies in 6 months

spoiler: most tools will not even survive 90 days

this guy is literally giving away the exact 2026 playbook for AI Agents.

he covers what to learn, how to build, and when to skip 👀

↓ read this today

[Ruben Hassid @rubenhassid]

Instead of watching Netflix tonight.

Spend a day mastering Claude here: https://t.co/Vn60ElPZ2i

→ Level 1 - 24 min: The basics.

Claude For Dummies: https://t.co/jw2qdIcjnh
Claude Certified: https://t.co/9jKsXWOt66
Stupid simple Claude: https://t.co/SVGd967eMQ

→ Level 2 - 1 hour: Real workflows.

Claude Cowork: https://t.co/uWTpOI3Woc
Claude for teams: https://t.co/qxlcqhf8bM
Claude Design: https://t.co/ZY8Fg5D2ea
Cowork + Projects: https://t.co/Q7AN9CZAbO
Claude for slides: https://t.co/L0bPMgXci6
Claude Skills: https://t.co/6cHYYfjXEA

→ Level 3 - 3.5 hours: The pro moves.

Claude to sound like you: https://t.co/kDGBpSF7Wh
Stop hitting Claude limits: https://t.co/j5fEzSH5br
Stop Prompting: https://t.co/j1LATSJiat
Claude replaced me: https://t.co/pNs1hPNDy5
Stop sounding like AI: https://t.co/JWKUGNKgOS
Excel with Claude: https://t.co/7g3CFNcKrs

→ Level 4 - 8 hours: Expert mode.

Claude Code: https://t.co/UgE9xBXVbE
Claude Connectors: https://t.co/TSAQqOpDeV
Stop using Claude at work: https://t.co/c6X55Thy6t

Pro tip: Don't binge it. Do one level per sitting.

Actually apply each guide before moving to the next.

[NeilXbt @neil_xbt]

Claude Cowork has real access to your real files.
Which means its mistakes have real consequences!

One early adopter lost 15,000 family photos to it.

That is not a reason to avoid the tool.

It is the reason to build one habit before anything else: back up important folders before running Cowork on them, and always make it show you the plan before executing.

Here is what that workflow actually looks like:
> You describe the outcome. Claude plans the steps.

> You review and approve. Claude executes. Finished work lands directly in your file system.

> Synthesize three months of client meeting notes into a single formatted report, organized by client, flagging action items mentioned more than once.

> Pull the last five project status reports from Drive, draft personalized email updates for each project lead, save them to Gmail, ready to send.

Available starting at $20 a month on Pro.
Both macOS and Windows.

The capability is real. So is the risk.
Treat it accordingly.

Read this article to get the best knowledge on using Claude Cowork!

Follow @neil_xbt for more Claude tooling intelligence that shows you the honest tradeoffs, not just the highlight reel.

[Anatoli Kopadze @AnatoliKopadze]

Karpathy just said the people who don't use LLMs are already losing

he spent 4 minutes explaining why smart people are still going to fall behind

not only the people who refuse AI, but also those who think signing up for Claude counts as using it

typing a prompt and reading what comes back isn't the skill, anyone can do that

the real shift is going from asking AI things to building something that runs without you

that's exactly why I wrote an article on Claude features most people don't even know exist

read the article below and you'll already be ahead of 99% of people

[0xTria @0xTria]

CHINESE CREATOR BUILT AN IPHONE APP THAT TELLS YOU IF A WATERMELON IS RIPE AND THE SCARY PART IS HE BARELY WROTE CODE

he opens Xcode, gives an AI coding agent a real SPEC, and the agent builds the app around one simple flow:

camera → analyze → recognition → result

the important part is not the watermelon

it's the workflow

prepare the idea
plan the database, server, payments
let the agent build in Xcode
test on a real iPhone

Supabase for data
Vercel for server
Zpay or Stripe for payments
multimodal AI for UI references

then he just keeps giving tiny fixes:
change the button
improve the cards
fix the bug
clean up the UI

this is what people still miss about AI coding

the app idea can be stupidly simple
the SPEC has to be clear
the agent does the boring part

[Lagosaur @0xLagosaur]

before you publish anything,

run it through one role: a red-team

paste this:

"attack this like a skeptic who wants me to look wrong

find the weakest claim and the missing caveat

name what an expert would push back on

return the 3 strongest objections and a one-line fix for each

do not soften them"

it finds the hole you talked yourself out of seeing

you ship the version,

that already survived its worst critic

[beamnxw ./ @beamnxw]

I USED TO WASTE HOURS ON CLAUDE. NOW I RUN A TEAM

The change? Building a real system instead of random prompts

Start with /init and CLAUDE. md. Add status monitoring, sub-agents, skills and worktrees. Reach agent teams on VPS

All 32 exact hacks from beginner to power user are quoted below with commands and real use cases

If I could do it, you can too. Read the guide.

[Codez @0xCodez]

Anthropic research lead:

"99% of our engineers are running swarms of 300+ self-improving agents.

close the agent loop. Give the model a way to verify its own output"

in a 20-minute session, Anthropic team member explains how to build a model that improves itself.

Claude + loops + plan mode + dynamic workflows -that’s the secret.

Watch the talk, then save the playbook below.

[NeilXbt @neil_xbt]

One German sociologist built 90,000 connected notes and wrote 70 books from them!

The notes were not the asset. The web of connections between them was.

Niklas Luhmann did not consult individual notes when he sat down to write. He followed the chains of connection, and the connections revealed implications no single note contained alone.

Most note systems get less useful over time. By month six, hundreds of notes you cannot find, a tagging system you abandoned, and a vague sense that all this captured information is doing nothing for you.

A knowledge graph inverts that trajectory entirely. The tenth note connects to nine existing notes. The hundredth connects to ninety-nine. Each addition makes every previous note slightly more valuable.

The hard part historically was the connection making. Manual labor most people abandoned before reaching the density where compounding begins.

Claude eliminates it. Reading across the entire vault simultaneously and finding the relationship between a note from January and a note from June that nothing in a folder system would ever connect.

Most practitioners report meaningful connections begin emerging after 50 to 100 connected notes.

The slow start is the network reaching density.

The complete build with the CLAUDE.md and atomic note format is documented here.

Bookmark it so you do not lose it!

Follow @neil_xbt for more Obsidian and Claude builds that show you what compounding knowledge actually looks like at scale.

[Movez @0xMovez]

Anthropic Product Lead:

"At Anthropic, our engineers are running swarms of 300+ agents daily.

Give your agents 100+ tools - just don’t load them all into context."

In a 30-minute talk, the Anthropic team shows how to deploy agents to production.

Claude + loops + routines + dynamic workflows - that’s the secret.

Watch the talk, then save the playbook below.

[RDeni @rostikdeni]

YOUR $20 CLAUDE SUBSCRIPTION CAN FEEL LIKE A FULL DEV TEAM

The difference is the moment Claude stops entering every task as a blank chat and starts with a process already loaded

A good Skill can carry the rules, examples, review standards, scripts, templates, and the exact way a task should be approached before you even explain what you need

That' s why the same Claude can feel completely different, one session is just an answer machine, another session feels like an architect, reviewer, debugger, technical writer, and operator sitting behind the same interface

The article below breaks down 50 Skills worth installing ⬇️