# dk32 live lane stopped (exec, 23 Oct 2026)

**Cause: every live dk32 buy was drained. Tokens left our wallet to 0x...dead within ~27 blocks of the buy, moved by a third party, not by our bot.**

Live record (fmdk32, $10 each): 6 buys filled, 5 closed at 0.000 ETH (4 stop/time, 1 "tp_3x" that only looked like 3x on the paper price), 1 still open_unsold. Also 6 buyfail (submitted, no fill) and 3 low_funds.
- Closed 5: wallet balance of each token = 0 at sell time, bot booked "no tokens left in the wallet" (out 0). On chain for 0x7d0361ec and 0x3849c167: Transfer wallet -> 0x000...dead, 27-29 blocks after our buy, tx sent by 0x617b0f55176a... via router 0xd6d35b56..., (multicall, Approval + Transfer logs). So the kit contract (or its operator) burns buyers' tokens right after the buy.
- Open 6th (0xadd84836...): bot sells fail 4x "GEvmValueLossTooHigh", kill switch fired 18:16:43 BKK, LIVE_DK32 removed by the bot's own kill. Tokens still in the wallet (36490.9e18); either burned later or unsellable. No restart, no change made.
- Loss: ~5 x $10 booked, plus $10 at risk on the sixth. Paper deadkit32 never models this (paper fills at quotes, no holding period in the wallet), so paper A 1.4x is not reachable live.

Fix options (all need a decision, none done by exec): (1) do not re-enable LIVE_DK32; retire the live lane. (2) If anyone wants it back: sell within the first ~25 blocks is not possible with 60 s dead/stop logic; a burn check (token transfer approvals, contract has burnFrom/operator hooks) would be needed in the rug gate, plus a replay of how many paper dk32 picks show the same burn.
Rug gate gap: pulled_before + round-trip sim cannot see a post-buy burn from the buyer's wallet. Suggest adding "holder balance still intact 60 s after a tiny test buy" or reading Transfer-to-dead of previous buyers of the same kit.