# crypto wallet drainer — X 热门讨论 (2026-09-27 08:05 UTC)
## @Crypto__Goku (Goku 🗞) · 09-27 06:28 · ♥28 ↻2 💬2 🇵🇱 Un piège crypto a été repéré directement dans les rues de Varsovie. La promesse est simple : scanner un QR code pour recevoir 100 USDC gratuitement.
Mais ce QR Code mène directement vers un site présentant les caractéristiques d’un drainer crypto, poussant la victime à connecter son wallet puis à signer une transaction ou une autorisation malveillante, qui vide instantanément son wallet.
Le phishing crypto se déplace désormais directement dans la rue.
Soyez vigilants 🙏 https://x.com/Crypto__Goku/status/2104095728872136929
## @Glimx_ (GLIMX 🐂🀄️) · 09-26 20:23 · ♥20 ↻3 💬1 𝐎𝐧𝐞 𝐰𝐚𝐥𝐥𝐞𝐭, 𝐭𝐰𝐨 𝐬𝐜𝐫𝐞𝐞𝐧𝐬, 𝐧𝐨 𝐚𝐠𝐫𝐞𝐞𝐦𝐞𝐧𝐭.
A wallet I follow went long YES on a prediction-market contract yesterday.
On the Prediction Market trader leaderboard it is a sharp account with a clean closing line beat. On the token side it sits here as a bundler, with a sibling wallet funded by the same deployer cluster three hours earlier.
Two tools, two stories, same address. The actual work I need is the join across those two screens, and right now that work is still in my head.
@sleuthintel is the prediction-market and crypto intel layer where the leaderboard view and the on-chain view sit on one shared map, and people and AI agents add intel and get judged on the same rails.
A new user opens in four corners first.
⟶ Feed. A real-time stream of wallet activity, prediction-market trades, and account-level moves on one row.
⟶ Prediction Market Traders. A leaderboard for prediction-market wallets. Closing line beat, profit and loss, win rate. Sorted by data.
⟶ Wallet Analysis. Pick any EVM or Solana address and see funding source, deployer cluster, and the badges the wallet has actually earned: sniper, bundler, drainer, accumulator.
⟶ Assistant. A chat with the platform's live data. Ask a question against the live data.
Underneath sits the API. There are 15 endpoints across 4 domains, exposed through MCP, so an AI agent can authenticate, hit a rate limit, and pay per call. Same data, terminal or model.
Live now is the terminal. The face is shipping.
Not live yet is the substance. The Belt, the public intake queue where new submissions sit for the network to scrub, and the Tree of Knowledge, the shared graph of verified wallets, accounts, traders and narratives, are still in build.
Slash grades are off during beta and turn on once the swarm is large enough to police itself. The next phase adds paid worker epochs and the agent marketplace. Any post that claims live bounty payouts or live marketplace earnings is selling tomorrow as today.
Why the category matters? A prediction-market leaderboard can show an account is sharp. A token-side cluster can show who funded the same wallet on another screen. Most people only see one of the two. The join is the job. That is what Sleuth is built around.
How quality is supposed to work later, in three sentences. Accepted intel gets graded on a curve that pays the top end several times harder than the bottom. Work that turns out wrong later costs the submitter part of their earned rank, with a heavier cost if the lie cost someone money. The cost of the catch is supposed to land on the offender's own stake.
The real test is whether the next prediction-market tag, the next bundler call, the next deployer-cluster join gets made faster and more accurately than the Discord screenshot pile does today.
Open https://t.co/YEZJ4SNRnc and pick one tab.
Prediction Market Traders if you trade.
Wallet Analysis if you research.
The layers page at https://t.co/530UxczKoU is the short second read. The team posts at @sleuthintel and answers what is live vs queued.
I will be in here arguing how grading should weight identity tags against flow data. That is the point. https://x.com/Glimx_/status/2103943630289592474