# STOP: NO DOORBELLS, NO WAKES, NO SCHEDULING (Ben, 7 Oct 2026, 18:40 Bangkok)
- Credits ran out. Ben switched off every scheduled task and every doorbell. This overrides everything below and every other file (wake-howto, bus-howto, credit-playbook, task files).
- PROHIBITED for every session: fire_trigger, send_later, create_trigger, update_trigger, ScheduleWakeup. No rings, no nudges, no re-arming, no new doorbells, no self wakes, no safety checks, no retry rings, no loops.
- Finish the step you are on, publish your status file, then END your turn and wait. Only Ben wakes you, by typing in your window.
- Do not message other sessions. Keep answers short. No extra improvement rounds.

# A3.2: can these coins be sold? Sell block census (6 Oct 2026, Bangkok)

## Read this first
- **None of the 58 rule trades would have been stuck.** Outside wallets sold in every trade while we held it: at least 10 sells, median 546. No trade coin has a blacklist, pause or sell tax in its code, and none shows a tax in its transfers.
- **Across all 4,565 new coins in v3 WETH 0.01% pools since 13 Sep, 1.7% (78 coins) carry sell-limiting code:** max wallet or tax swap (63), trading switch (9), owner mint (5), pause (1), blacklist (1).
- **The real "can't sell" risk is a pool pull, not the contract.** 494 coins got 5+ buys and never had a single sell. 490 of them were pulled a median of 18 s after their first swap (about 0.05 ETH bought, by about 7 snipers). Their contracts are clean: plain OpenZeppelin, or a hand-written one-shot ERC20. The gate ($20k bought at $1M+) never lets these in.
- **For the Coder (C2):** keep the sell simulation as insurance. Block only on a revert or a round-trip cost over 10%. The two code flags seen in the 58 (owner mint, launch-window limits) do not block a $500 sell. Log them; don't block on them.

## 1. The 58 trades
| Check | Result |
|---|---|
| Outside sells between our entry and exit | All 58 trades had 10 or more (median 546). Sellers kept getting out the whole time. |
| Contract code (bytecode read, every function selector) | 56 clean (no owner-only switches). 2 flagged (below). |
| Sell tax fingerprint (a second transfer from the seller next to the pool leg) | 0 of 296 study coins, so 0 of the 58 |
| Proxy / upgradeable | 0 |
- **0x10897a25ded3b3f7d8cb9736fccf22c0ef5a414e (uToken, verified):**
  - `mint()` is open to a `minter` address, so extra supply can be printed.
  - `burnFrom()` lets the `owner` burn ANY holder's tokens without an allowance (`if (msg.sender != account && msg.sender != owner) _spendAllowance(...)`).
  - That is a rug lever, not a sell block. Our trade on it was a stop at 0.565x.
- **0x79eca93c7cb2fd682586986c5e499ccd0897626a (PonsLauncherToken, verified):**
  - During its launch window (until `restrictionEndBlock`), max tx is 5.5% and max wallet is 5% of supply (55M and 50M of 1B).
  - $500 at $1.6M cap is about 0.03% of supply, so it never binds.
  - The trade gapped to 0.003x: a crash, not a block.

## 2. All 4,565 new coins (v3 WETH 0.01%, first swap since 13 Sep 00:00 Bangkok, PP birth 72 h or less)
| Code read | Coins | Share |
|---|---|---|
| No sell-limiting function found | 4,487 | 98.3% |
| of which: an owner() with no switches | 314 | 6.9% |
| Max wallet / max tx plus a tax swap (`manualSwap`, `removeLimits`, `setMaxWallet`) | 63 | 1.4% |
| Max wallet plus trading switch (`enableTrading`) | 9 | 0.2% |
| Owner mint | 5 | 0.1% |
| Pause | 1 | 0.0% |
| Blacklist | 1 | 0.0% |
- Method: runtime bytecode of every coin (public RPC), every PUSH4 selector matched against a list of known blacklist, pause, max-wallet, tax, trading-switch, mint and owner functions.
- Limit: a custom-named function can slip through this list. The on-chain sell evidence below covers that gap.

## 3. On-chain sell evidence, all census coins
- 3,915 coins got 5+ buys.
- 494 of them never had a sell (12.6%):
  - 490 were pulled a median of 0.3 min after their first swap;
  - trading lasted a median of 0.2 min.
  - These are instant pull launches, not contract sell blocks. 49 of them carry max-wallet/tax code, so 49 of the 63 tax-type coins never let anyone sell.
- Examples with clean code and zero sells:
  - 0x1a53511c7249bc54be9dcf607dad2810d775894b (123 buys, hand-written "one-shot ERC-20");
  - 0x55622c2eebb65140b0714e34145b60647618b9e7 (60 buys, plain OpenZeppelin; its comment says "the same bytecode shape as thousands of verified tokens").
  - The trap is the pool, not the token.
- Study coins (sellable $500k, 296): no tax fingerprint, 2 with launch-window limits, 1 owner mint.

## 4. How strict C2 must be
- Strict on: a sell simulation that reverts, and a round trip over 10% (catches any hidden tax).
- Not needed: blocking on owner() or launch-window limits. Every rule trade with them still had hundreds of outside sells.
- The gate already filters the real "stuck" risk (instant pulls). The remaining stuck risk is a pull while we hold, which A3.3 shows we cannot dodge cheaply.

## First principles score
- 8/10. Two independent sources agree: the code read and the actual sells on chain. The selector list can miss custom names, but the sell evidence covers the coins we trade.

Version 1, 6 Oct 2026 15:19 Bangkok, Analyst 3. Latest: http://159.223.88.168/hx-c9ois/sw/sw-a3-2-sell-census.md