tuanta✸,✸

ANTHROPIC ACCIDENTALLY LEAKED Claude Code source code and someone made a copy of it on GitHub, in just 1 hour it already got 1.5K stars.

Let your agent orchestrate thousands of sub-agents and build harnesses: github.com/shouc/agentflow Codex/Claude Code/Kimi CLI supported.

Claude code source code has been leaked via a map file in their npm registry! Code: …a8527898604c1bbb12468b1581d95e.r2.dev/src.zip

Claude Code + computer use is f*cking cracked 🤯 Build a landing page → Claude opens Chrome, looks at it, spots every issue, and fixes it — without you describing a single thing. All inside Claude Code. Perfect for DTC brands and agencies who are still vibe-coding landing pages and advertorials in Claude Code, then manually opening them in Chrome, spotting 15 things wrong, and describing every visual issue back to Claude one at a time. If you're building pages in Claude Code and your workflow looks like this — build the page, open it in Chrome, spot broken spacing, go back to Claude, type "the CTA button is too low and the hero image is cut off," wait for the fix, open Chrome again, find 3 new issues, describe those too ... Claude Code + computer use eliminates the entire loop: → Claude writes the full landing page or advertorial → Opens Chrome and navigates to it → Spots layout issues, broken spacing, off-brand colors, missing elements → Fixes everything and re-checks until the page looks right → Tests your Shopify product pages by clicking through like a real customer → Walks through your checkout flow and flags friction before customers hit it → You only see the finished, visually verified result No describing what you see on screen. No "the CTA button needs more contrast" back-and-forth. No being the eyeballs for an AI that can't see. What you get: → Landing pages and advertorials Claude builds AND visually QAs before you ever look at them → Product pages Claude clicks through — testing layout, images, and CTAs like a real user → HTML dashboards Claude opens and verifies the charts actually render → Checkout flows Claude walks through step by step to catch friction → All of it happening in one session — build, test, fix, done One prompt. Claude builds it, checks it, and fixes it. You just review the finished page. I put together a full playbook with the exact setup, the prompts, and 5 DTC workflows that use Claude Code + computer use. Want it for free? > Like this post > Comment "CLAUDE" And I'll send it over (must be following so I can DM)

CLAUDE BUILT A BOT THAT TRADES LIKE A MARKET MAKER AND PAYS ITSELF FROM BOTH SIDES OF THE BOOK. While most traders try to guess the winner, this one just sits in the middle, captures the spread, and keeps climbing.

Stop copy and pasting prompts one by one Save this article and thank me later x.com/kloss_xyz/stat…

CLAW3D JUST MADE MANAGING AI AGENTS A LOT EASIER. You can now add and remove agents directly inside the interface without jumping back into your OpenClaw setup.

MIROFISH TURNED CLAUDE + QUANT FORMULAS INTO A 274 AGENT POLYMARKET TERMINAL THAT MADE $15,473 IN 14 DAYS. It scans contracts, runs Bayes updates, calculates EV, sizes with Kelly, and trades what the math says instead of what feels right.

ANDREJ KARPATHY JUST DROPPED “AGENTHUB” ITS A 100% OPEN-SOURCE GITHUB REBUILT FOR AI AGENTS Github: github.com/karpathy/agent…

KV caching, visually explained: (how and why it works)

OpenClaw meets RL! OpenClaw Agents adapt through memory files and skills, but the base model weights never actually change. OpenClaw-RL solves this! It wraps a self-hosted model as an OpenAI-compatible API, intercepts live conversations from OpenClaw, and trains the policy in the background using RL. The architecture is fully async. This means serving, reward scoring, and training all run in parallel. Once done, weights get hot-swapped after every batch while the agent keeps responding. Currently, it has two training modes: - Binary RL (GRPO): A process reward model scores each turn as good, bad, or neutral. That scalar reward drives policy updates via a PPO-style clipped objective. - On-Policy Distillation: When concrete corrections come in like "you should have checked that file first," it uses that feedback as a richer, directional training signal at the token level. When to use OpenClaw-RL? To be fair, a lot of agent behavior can already be improved through better memory and skill design. OpenClaw's existing skill ecosystem and community-built self-improvement skills handle a wide range of use cases without touching model weights at all. If the agent keeps forgetting preferences, that's a memory problem. And if it doesn't know how to handle a specific workflow, that's a skill problem. Both are solvable at the prompt and context layer. Where RL becomes interesting is when the failure pattern lives deeper in the model's reasoning itself. Things like consistently poor tool selection order, weak multi-step planning, or failing to interpret ambiguous instructions the way a specific user intends. Research on agentic RL (like ARTIST and Agent-R1) has shown that these behavioral patterns hit a ceiling with prompt-based approaches alone, especially in complex multi-turn tasks where the model needs to recover from tool failures or adapt its strategy mid-execution. That's the layer OpenClaw-RL targets, and it's a meaningful distinction from what OpenClaw offers. I have shared the repo in the replies!

another dope pixel office for agent swarms, honestly these keep getting better and better, and the idea of watching your agents work in a mini game instead of staring at logs just hits different discord and telegram are great for talking to your agents individually but having them all in one mini game with a minimal control panel where you can just look at the screen and know the state of your entire swarm / what theyre working on is something else we are going to see a lot more of this in the coming weeks, visual layers on top of agent infrastructure... it will not replace the bottom layer ofc but it would make it feel alive, like you are actually running a team the moment you can visually manage your agents, everything changes, you spot what is out of sync, what stopped working, and what needs attention way easier.. credits @meng_shengyu for building it.

Orchestration framework for AI agent loops github.com/mikeyobrien/ra…

OpenClaw can now scrape any website without getting blocked - zero bot detection, bypasses Cloudflare natively, 774x faster than BeautifulSoup. No selector maintenance. No workarounds. Just data. THIS IS AN UNFAIR ADVANTAGE AND IT'S FULLY OPEN SOURCE.

this guy just hired 25 founding engineers (it's 25 sessions of claude code)

@0x_Discover How to sir

I built my own script for 5-minute BTC markets using ClawdBot on Polymarket. Still can’t believe it - $20 turned into $1,539 overnight. No insider info. No Elon connections. Just sat down and wrote the script. Copytrade -t.me/PolyGunSniperB… And honestly — it’s not “rocket science.” No massive datasets. No insane infrastructure. Just clean logic + execution. Full framework: 5-minute BTC & ETH micro-arb The system targets short-cycle contracts. When YES + NO briefly price below $1, it enters both sides and locks the spread. Moltbot wired directly into Polymarket executes instantly. No forecasting. No bias. No discretion. Speed > hesitation When volatility spikes and manual traders pause, the script fires. No emotion. No delay. No second-guessing. By the time others react, the gap is gone. Automation compounds small edges Each trade captures cents, not home runs. But high-frequency repetition stacks small spreads into meaningful returns. Scale is the multiplier 8,894 trades. Tiny individually. ~$150K collectively. At that point it’s not about prediction. It’s about infrastructure. Build → deploy → let it run.

$400K in under a month. This OpenClaw setup is averaging: • ~$5 per second • ~$300 per hour • ~$7K per day Proof: @0x0eA574F3204C5c9C0cdEad90392ea0990F4D17e4-1769515653156" target="_blank" rel="nofollow noopener">polymarket.com/@0x0eA574F3204… Copytrade → t.me/PolyGunSniperB… And it kept running even after Polymarket removed the 500ms delay. What it does: • Trades only 5-minute BTC Up/Down markets • Buys YES + NO repeatedly in the first ~4 minutes • Enters when combined price drops below $1 • Locks the spread before expiry 6,823 trades. Pure arbitrage structure. No directional bias. It compounds small gaps over and over - scaling size as balance grows. Not prediction. Not narratives. Just pricing inefficiencies + automation.

Yet another accurate prediction.