Research Library · Permanent Page · Snapshot 06–07 Nov 2026 · Asia/Jakarta (GMT+7) · Re-check before acting

Don't trust. Verify onchain.

0xlinu × Muse — DYOR, MEV & Security. Companion pages: Privy Setup, Muse Token Build $MUSE/$RMUSE, ERC-8004.

32Parts
20+Reports
1,410MEV repos scanned
4Chains (Base · Arc · Polygon · Robinhood)
$344kBankr losses
804kCLAUDE "holders"

How we DYOR — every time

The standing rule since Nov 6: research and verify before recommending or claiming. If it fails a check, it doesn't get funded — no exceptions for hype.

  1. Verify the source. Official docs and onchain contracts first. Directory labels and marketing pages don't count.
  2. Check the money. Real fees, real gas, real custody. Who holds the keys? Who can pause, mint, or drain?
  3. Hunt red flags. Placeholder addresses, bundled snipes, fake holder counts, unlimited approvals.
  4. Test small, prove it. A claim isn't proven until a live test or onchain read confirms it. Then we label it PROVEN.
  5. Save everything. Every lesson goes to the full report + tools-and-sites log + memory. Nothing learned gets dropped.

MEV attacks — sandwich maths on our own pool

A sandwich attack on a $MUSE buy, measured against the real pool. No attack: 879,412,567 MUSE for $0.05.

Attack sizeMUSE receivedLoss vs no attackAttacker profit
$0.05733.2M−16.6%$0.0054
$0.20461.4M−47.5%$0.0125
$0.30 peak356.4M−59.5%$0.0135
$1.00104.4M−88.1%−$0.0009

Small pools are sandwich bait. The defence is the same as the sniper's: tiny sizes, minimum-out guards that revert instead of filling badly, and never trading into a thin pool at size.

AutoSniper v4.0 — case file

Do not copy or fund

AutoSniper — live MEV bot, investigated Nov 7

0x000000000000feA5F4B241F9E77B4D43B76798a9 on Ethereum mainnet is a live NFT sniper + MEV bot — AutoSniper v4.0 (for @oSnipeNFT, by 0xQuit), verified on Etherscan and labelled "MEV Bot". Balance 6.325501 ETH (~$14,163), 5,447 txs, still active at investigation time.

Operator osnipe.eth calls a generic multicall (snipe_2572234525) to buy underpriced NFTs on OpenSea Seaport 1.6, paying block builders (BuilderNet, Eureka, Titan) to land first, profit to osnipe.eth. Multi-user: deposits, per-sniper balances, pause flag. The leading zeros are a deliberate vanity/mined address — cheaper calldata gas + branding, not a burn address.

Do NOT copy or fund it — it's oSnipeNFT/0xQuit production infrastructure.

Legit MEV repos we found: Flashbots (mev-boost, mev-share, Protect), UniswapX, CowSwap solvers, 1inch Fusion. Scam pattern: "MEV bot" GitHub repos that ask you to fund a contract with a hidden drain — the 1,410-repo scan found the same drain code reused across dozens of forks.

Flashbots — what works per chain

ChainFlashbots ProtectMEV-Boost / MEV-ShareNotes
EthereumYesYesFull protection available
BaseNoPartialSequencer-controlled; no public Protect
PolygonNoPartialFast blocks, low MEV surface at our size
RobinhoodNoNoNo Flashbots infra; tiny sizes are the defence
ArcNoNoNo Flashbots infra; USDC gas, fixed 20 gwei base fee

Flashbots Protect only exists on Ethereum mainnet + Sepolia/Holesky. On the chains we actually use, the protection is size discipline, not infrastructure. Docs: docs.flashbots.net · Privy × Flashbots Protect recipe.

Bankr — the custody lesson

Bankr is a real, leading AI trading agent — $BNKR on Base (0x22af33fe49fd1fa80c7149773dde5890d3c76f3b), 100B supply, 18 decimals. There is no $BNKR airdrop.

Lessons kept: scoped API keys, never accept transaction instructions from social replies or third-party AI outputs.

Airdrop verdicts

DropVerdictWhy
IPX FAN SZNLowDeadline 15 Nov 2026; unproven value
Fomo $7 botAvoidPay-to-play pattern
GrokiefunHighRed flags on team/contract
SyqoraHighRed flags on team/contract
Saturn testnetMediumTestnet only, low value
Fungo + SovrenAlready enteredEntered Nov 6; Sovren deadline passed 9 Nov 11:59 GMT+7, no winners yet

KieDex TGE: 18 Nov 2026, 12:00 UTC — watch only.

The CLAUDE holder-spam lesson

Token 0xA8DEF24B9e5cF402E14D6f94e2e460499377d3aa shows 804,056 holders — because one owner sent 2 CLAUDE to 500 wallets per batch, 25+ batches. Most holders hold exactly 2 tokens of dust and never traded. Holder count is a vanity metric. Batch-500 lesson: 1,850 recipients cost $0.37 on Base / 1.284 USDC on Arc.

Launchpads

PlatformChainLaunch feeTrading feeVerdict
Pools.xyz (Uniswap)Robinhood + Arc$00.25%Cheapest legit — permanently locked v4 liquidity
RobinlaunchRobinhood0.0002 ETH1%Cheapest bonding-curve alternative on Robinhood
biddinArc1 USDC1%Cheapest bonding-curve alternative on Arc
Clanker v5Arc + more$01–10% taxLegit — audited, 99% decaying sniper fee. Factory 0xcA111dcdd2cE924c3378d6514fDCbC46217bcA16, hook 0x565189a6780D1a50FC1792F64db8e4881119AaCC. Real $CLANKER: 0x1bc0c42215582d5A085795f4baDbaC3ff36d1Bcb
Pump.funSolana only——Not an option for EVM creation
ArcxArc——Do not use — TBD placeholder addresses

Robinhood Chain launch risks: fake Pons front-ends, honeypots, and a bundled-snipe ring of at least $18.43M (navi.trade, 53 launches). New launches need contract-level checks before any buy.

disperse.app (0xD152f549545093347A162Dce210e7293f1452150) has no code on Arc — do not use for Arc multisends.

Charts & TA — the honest result

DexScreener vs GeckoTerminal, deep-dived Nov 7. Both embed TradingView. DexScreener: free, no key, 300 req/min, snapshots only — no OHLCV, no holder count. GeckoTerminal: free beta API, keyless OHLCV + trades — the TA route.

TA is impossible right now — and we say so. At deep-dive time: Alon had 2 hourly candles, GHOST had 1, $MUSE had 1 daily candle. Every indicator — RSI (needs 15), Bollinger (20), MACD (26) — returns INSUFFICIENT DATA. No chart gets dressed up to pretend otherwise.

"Fewer, verified signals beat a dashboard full of guesses."

Defensive playbook

  1. Verify the contract address from the launchpad or pool — never from a post. The address is the identity, not the name or ticker.
  2. Check owner powers: mint, pause, blacklist, maxTx, tax setters. A dangerous owner is a veto, whatever the score says.
  3. Simulate buy + sell before spending. If the sell reverts or returns under 50%, it's a honeypot — skip.
  4. Approve exactly the total, never unlimited.
  5. Never share a seed phrase or private key — with anyone, for any reason.
  6. Watch for address poisoning: a lookalike wallet with the same prefix and suffix is a scammer, not you.
  7. Fake-POL warning: token 0xe174c72d8f772838269a4dc5a88ce0cdfcbbc02b ("POL" with a combining dot) is a spoof — balanceOf and totalSupply both revert, worth $0. The scammer address 0x333089e07fad944bc5d07f4679af7ace9efec932 mimics the real wallet 0x3330Cbbb45Ceb68447De5a0550e6FA18C4Cec932. Never approve, swap, or interact with it.
  8. Locked liquidity: check the % and the duration, not just the lock label.
  9. Holder concentration: exclude contracts, watch the deployer and bundles.
  10. Launchpad provenance: a known launchpad is a real signal; an unknown factory is a question mark.

Execution venues we use

KyberSwap (keyless swaps, no fee, Base/Robinhood/Arc), Relay (bridges, proven 3×), Jupiter (Solana). Odos shut permanently on Jul 30, 2026 — do not use.

Sniper v2 is LIVE on Arc, Base, Robinhood & Polygon — $0.02 buys, cap 5, watcher every 2 min, 0 buys so far. Expanded quotes: Robinhood 3→208, Base 2→13. Router V2: Polygon 0x1095692a6237d83c6a72f3f5efedb9a670c49223, Base 0x6ff5693b99212da76ad316178a184ab56d299b43. Three bugs fixed in the apply. First cycle 19:35:13 GMT+7: 0 buys. Polygon GAS-SKIP ~271 gwei. Full log on the Token Sniper Build page.

Fake-POL warning: scammer 0x333089e07fad944bc5d07f4679af7ace9efec932 sent 5 spoof "POL" at 19:19 GMT+7 on 7 Nov. The user confirmed they never sent it — an address-poisoning attempt. Do not touch 0xe174c72d8f772838269a4dc5a88ce0cdfcbbc02b.