CrprtoWIN retweeted
1/
TermMax is launching a Booster program on @BinanceWallet.
Complete five tasks or post on Binance Square to share 2,000,000 $TMX.
Aug 17 07:00 UTC to Aug 24 23:59 UTC
$MU It's a US based company 😺
Season 5 of Agent Arena is in the books. US stocks + commodities edition.
Here's who came out on top:
🥇 @wincrypt_ — MU Extreme Reversion, +$4,099 PnL
🥈 @dayanBTC — Ultra-Fast Bollinger Reversion, +$1,536 PnL
🥉 @ISchuwenck — Extreme Short - Bollinger Strategy, +$1,502 PnL
Big respect to everyone who deployed this round.
Build your own: moss.site/agent
CrprtoWIN retweeted
ANTHROPIC JUST LOST 20,000 CLAUDE CODE USERS TO ONE GITHUB REPO.
20,000 devs. 26k stars. 4,000 forks. 10 free models. COMPLETELY FREE.
Bookmark this before Anthropic notices. Your Claude Code bill will start working differently.
It reroutes Claude Code to DeepSeek, Kimi, and 8 more same CLI, different model behind it.
5 min setup. $200/mo saved. 0 downgrade on small tasks.
swap the endpoint in your config today.
Link below.
Claude Code → Endpoint swap → Free models → No bill → Money
CrprtoWIN retweeted
ZachXBT on TG just said all hardware wallets are garbage and using a separate iPhone is safer..
i agree with the main lesson.. but a separate iPhone alone still leaves you with one device and one seed as the single point of failure.
for serious funds, this is the exact setup i would recommend:
1) Create 3 completely fresh signer wallets
- signer 1: dedicated iPhone
- signer 2: separate phone or hardware wallet
- signer 3: backup device stored somewhere else
never use your everyday wallet as a signer.
2) Lock down the dedicated iPhone
- no SIM
- no email, Telegram or social apps
- no random browsing
- install only the official wallet app
- use a strong passcode and Face ID
- enable Stolen Device Protection
- use it only when you need to sign
write the recovery phrase on paper or steel and store it offline.
never screenshot it, save it in Notes, upload it to iCloud or type it into any website.
3) Create a 2-of-3 Safe multisig
manually visit app.safe.global, choose the correct network and add the 3 public wallet addresses as signers.
set the threshold to 2 out of 3.
this means any transaction needs approval from 2 separate devices before funds can move..
you only add the public addresses. Safe will never need your seed phrases or private keys..
4) Test everything first
send a very small amount to the Safe.
then try sending it back by approving the transaction from 2 different signers.
only move serious funds after you know the entire setup and recovery process works.
5) Move the remaining funds in batches
don’t transfer your whole portfolio in one transaction.
send a small batch, confirm it arrived on the correct network and then continue.
6) Treat the Safe like a vault
don’t use it for random mints, airdrops or unknown dApps.
use a separate spending wallet for daily activity and keep only the amount you can afford to lose there.
before approving anything, verify the network, token, amount and receiving address on both signer devices.
@zachxbt
CrprtoWIN retweeted
Agent Arena S5 is open. Only US stocks & commodities.
July 7-16, 12pm UTC. Highest PnL wins.
🥇 3mo Claude Max + 1k💎
🥈 1mo Claude Max + 800💎
🥉 1mo Claude Max + 500💎
Each Participants: 200💎
Think you can trade equities better?
Build now 👉moss.site/agent
Coinbase has switched to Chinese open-source AI models like Zhipu GLM 5.2 and DeepSeek, slashing internal AI costs by nearly 50% while increasing usage.
GLM 5.2 now outperforms GPT-5.5 on key coding benchmarks like SWE-bench Pro (62.1 vs 58.6) at a fraction of the price.
Open-source models are closing in fast on closed-source leaders.
$eth to 0
NVIDIA and Jensen Huang have really been holding back something massive. This is fucking impressive.
opus 4.6，gpt -5.5 Is the best
so opus 4.8 is out, and the funny part is they didn't really chase the leaderboard.
scores barely moved past 4.7. the whole release is token efficiency, like ~61% cheaper to run and fast mode 3x cheaper.
that tells you what they were actually worried about
so the 4.8 vs gpt 5.5 question isnt whos smarter anymore.. they're too close for that..
its what are you using it for...so I've made a little cheat sheet for it
opus side: repo-scale coding, long unattended agents, work where it needs to catch its own mistakes, computer use..
codex side: terminal-heavy stuff, web research, high-volume pipelines, anything speed sensitive..
and theyre basically tied on pure reasoning + time to first token...
CrprtoWIN retweeted
I'm close to hitting 50K on X. Should celebrate.
Giving away 3 Pro subs to @CommandCodeAI
STEPS
1. Follow @MrAhmadAwais
2. Follow @CommandCodeAI
3. Repost this post
And reply with what you're your building with Command Code.
Will pick randomly soon as I hit 50K. LFG!
OnlyFans user data is currently being sold on dark web forums. A seller is offering what they claim is a database of approximately 340 million user records.
However, this is not the result of a direct hack or breach of OnlyFans' systems. The seller has admitted that they did not breach OnlyFans. Instead, they compiled the data by combining information from previous breaches of other platforms (such as X/Twitter, Instagram, etc.) with publicly available OnlyFans profile information.
The dataset reportedly includes usernames, emails, phone numbers (for some users), registration dates, subscriber/fan counts, and limited payment card details (last 4 digits).
While this is still a serious privacy concern — especially for creators — it is important to clarify that OnlyFans itself was not hacked in this inciden
OnlyFans is Hacked 🚨
Apparently OnlyFans has been hacked and they're selling the complete database of 340 million users
including data of content creators and consumers.
The leaked data includes
- Usernames and profile names
- Email addresses
- Phone numbers
- Account creation dates
- Follower/subscriber metrics
- Creator/fan rankings
- Linked social media profiles
- Partial payment card metadata (last 4 digits of the card)
The result of this is going to be a massive wave of extortion attempts against users
“She could’ve just lived off her acting fame, but Resident Evil star Milla Jovovich jumped into AI and open-sourced MemPalace — achieving the first perfect score ever on LongMemEval (500/500). Real tech democratization.”
My friend Milla Jovovich and I spent months creating an AI memory system with Claude. It just posted a perfect score on the standard benchmark - beating every product in the space, free or paid.
It's called MemPalace, and it works nothing like anything else out there.
Instead of sending your data to a background agent in the cloud, it mines your conversations locally and organizes them into a palace - a structured architecture with wings, halls, and rooms that mirrors how human memory actually works.
Here is what that gets you:
→ Your AI knows who you are before you type a single word - family, projects, preferences, loaded in ~120 tokens
→ Palace architecture organizes memories by domain and type - not a flat list of facts, a navigable structure
→ Semantic search across months of conversations finds the answer in position 1 or 2
→ AAAK compression fits your entire life context into 120 tokens - 30x lossless compression any LLM reads natively
→ Contradiction detection catches wrong names, wrong pronouns, wrong ages before you ever see them
The benchmarks:
100% recall on LongMemEval — first perfect score ever recorded. 500/500 questions. Every question type at 100%.
92.9% on ConvoMem — more than 2x Mem0's score.
100% on LoCoMo — every multi-hop reasoning category, including temporal inference which stumps most systems.
No API key. No cloud. No subscription. One dependency. Runs on your machine. Your memories never leave.
MIT License. 100% Open Source.
github.com/milla-jovovich/me…
Readers added context they thought people might want to know
The claimed 100% LongMemEval score uses targeted fixes for the 3 failing questions and LLM reranking (held-out score: 98.4%). The 100% LoCoMo score uses top-k=50 exceeding session count with reranking (honest top-10 no rerank: 88.9%). github.com/milla-jovovich…
ANTHROPIC JUST ACCIDENTALLY LEAKED THE ENTIRE CLAUDE CODE SOURCE (512,000 LINES) — And the Community Turned It Into Pure FIRE
Anthropic’s team shipped an npm update… and forgot to delete the debug source maps (.map files). Boom — the full TypeScript source code for Claude Code spilled out into the open.
What happened next is straight-up legendary.A Korean developer (@realsigridjin— the guy who burned 25 billion tokens last year) rewrote the whole thing overnight into claw-code (clean Python version).
The team is now rebuilding v2 in Rust… and the code is already permanently on IPFS. Anthropic can’t delete it even if they wanted to.
Another dev (@paoloanzn) recompiled it into free-code — the unrestricted, fully unlocked version.
This is the biggest open-source AI gift of 2026 Here’s exactly what you can grab right now:
1. claw-code (Python rewrite)
→ Local, clean Agent harness. Run Claude Code-style tools 100% offline. Perfect for studying architecture and building fast.
github.com/instructkr/claw-c…
2. free-code (The unrestricted build)
→ All telemetry & enterprise guardrails deleted. Every single hidden/experimental feature turned ON
github.com/paoloanzn/free-co…
3. Full Hidden & Unreleased Features Bible
→ Complete list of everything Anthropic never publicly showed: 8 unreleased features (Buddy AI pet, KAIROS, Ultraplan, Coordinator Mode, etc.), 26 secret slash commands, 32 build flags, 120+ secret env vars.
ccleaks.com
These are 100% community forks — unofficial, not affiliated with Anthropic. They may have bugs,
stability quirks, or legal gray areas. Use only for local experimentation and learning. Evaluate at your own risk.
Kraken & Legion’s first launch project
@sportfun has released its allocation rules.
FDV is $60M at TGE.
50% unlocks at TGE, with the remaining 50% vesting linearly over 6 months.
Effectively, this is closer to a $120M fully diluted setup — if it opens at 2x, you’re already breakeven.
Worth paying close attention to how allocations are handled this time.
Allocation rules on Legion
Up to 50% allocated to community members and long-term supporters — basically users who have previously interacted
Up to 20% allocated to high social-score, high-influence users (in plain terms: KOLs)
At least 10% allocated to Legion Diamond users and high-score accounts
At least 5% allocated to users with high scores but smaller application sizes
Allocation rules on Kraken
80% allocated to high-score users. The more criteria you meet, the higher your chances:
Kraken+ subscription
Account age
Trading and custody activity
FOMO multiplier (boost if you were not selected in previous sales)
HODLer bonus — holding tokens from previous Kraken Launch sales for 30 / 90 / 180 days
20% allocated via random lottery
Maximum participation per person: $25,000
This one likely has some upside. We’ve seen similar setups before, and while institutional interest, TVL, user count, and hype are all fairly average, it is still Kraken’s first launch under this structure, and the rules themselves are reasonably well-designed.
There are risks, of course.
DYOR.
