This is a test deployment with negative expected value: trades are ~$1 through very thin pools, and BDAI taxes the vault's trades 0.6% on every buy and 2.5% on every sell. The live vault trades the BDAI/WPLS venue; the direct BDAI/DAI pair (where the successor vault traded) is only ~$1.3k deep. Expect to lose value. Deposit only what you're happy to lose helping test. On top of that, BDAI carries token risks the BUSHI vault never had (last verified 2026-07-14):

1 · The token owner still holds the keys. BDAI is not renounced: the owner wallet 0xB68C…Dc78 keeps tax-exclusion powers over the token. There are no mint, pause, or blacklist switches, but this token can still be tuned by one person.

2 · The owner dominates the liquidity on BOTH venues, and has pulled it before. WPLS venue (the live vault's pool): the owner has historically held ~97% of the BDAI/WPLS LP, unlocked, with a history of pulling it. Direct BDAI/DAI pair (the successor vault's venue): roughly 92% of the LP is pullable: the owner (25.57%) plus an unverified staking farm (66.42%), none of it locked, including a same-day drain on 2026-07-13 (~59,880 BDAI net, routed through a wallet the owner funded). If either pool gets pulled, that venue's BDAI side becomes unsellable and effectively worthless; only the DAI held in the vault keeps its value.

3 · Every BDAI transfer consults an unverified third-party "Helpers" contract (plus its fee factory), both owned by one external party. It's the code that decides which transfers get taxed and how. Because it can't be read, it is a potential transfer-freeze vector: BDAI transfers (including the BDAI leg of your withdrawal) could conceivably be bricked.

4 · A one-way reflection trap. If that Helpers contract ever classifies the vault as a "pair", the vault is permanently excluded from BDAI reflections. Irreversible. (Reflections here are tiny anyway, but the door only swings one way.)

If any of this happens, the operator can pause trading and you can still withdraw whatever the token still lets the vault transfer, but nobody can restore value the token or its owner destroys. The vault contract itself cannot protect you from any of the above. Deposits and withdrawals are expected untaxed on BDAI, and the 24h cooldown after each deposit works exactly as on BUSHI.

RADX carries the strongest risk register Tax Haven has ever published. The operator reviewed the findings below on July 11, 2026 and explicitly accepted them. The reasoning: the vault takes on the same token-contract risk the RADX community itself takes on. Acceptance is a disclosure decision, not a safety finding. This is a negative-expected-value test deployment at ~$1 trades; deposit only what you're happy to lose helping test (last verified 2026-07-13):

1 · The token's source is only partially corroborated. TokenTax source became readable on July 12, but its optimizer metadata conflicts with the deployed metadata, so it is mechanic evidence, not exact-match proof. The surrounding factory, Helpers, and custody contracts are unverified, including two unverified upgradeable proxies that hold 80% of the total supply.

2 · An OPEN factory-mediated mint path. The token owner has no direct mint button, but the unverified owner-linked factory can initialize a minter while minting is unfinalized and then inflate the supply without a disclosed cap, including into the pool this vault trades against. The vault cannot prevent that.

3 · About 96% of the liquidity is removable, most of it by one party. The single RADX/WPLS pool held about $1,000 total (≈$503 per side) at review; ~59.5% of the LP sits in an unverified owner-created contract and another ~32% in ordinary wallets. One transaction can effectively drain the pool and make RADX unsellable; only the Ethereum-bridged DAI held in the vault would keep its value.

4 · One account controls the whole ecosystem. A single externally owned account owns the token (not renounced), the factory, and the custody stack. It can grant or revoke arbitrary tax exemptions, change tax receivers, and change the Liquify slice's token address, which is an arbitrary external-call surface inside transfers.

5 · The token sells its own tax inventory inside YOUR transfers, with zero minimum output. On every non-excluded wallet-sourced transfer (deposits, sells, withdrawals, plain transfers) RADX dumps its accumulated tax tokens into the pool with no slippage protection at all, making the embedded swap fully sandwich-able by anyone watching the mempool and adding gas to three of the vault's four legs. The fork rehearsal measured a dump 1.03× the size of the vault's own sell inside a single transaction. Buys never dumped on the fork.

6 · A token-side failure can block the whole withdrawal. A basket withdrawal sends RADX and DAI in one atomic transaction. A known-bug "sendWETH" receiver failure inside the token can brick later taxed transfers, and the owner-mutable Liquify token address can introduce an external call that reverts. If the RADX leg reverts, the DAI leg stays in the vault too. There is no DAI-only emergency withdrawal, and no pause, operator change, or website fix can override the external token.

7 · RADX taxes YOUR deposits and withdrawals: 5% each way. The token charges a measured, exact 5% on every ordinary transfer, including wallet→vault deposits and vault→wallet withdrawals. Deposit-then-withdraw keeps 90.25% before trading, AMM fees, price movement, the self-dump above, or gas even enter the picture.

8 · Pushed Ethereum-bridged DAI reflections are switched on and flowing. The token's tax machinery pushes DAI to holders; DAI pushed to the vault joins the shared DAI pool and is split pro-rata by shares — the vault's "DAI reflected" stat above tracks the running total. It is not a separate reward you claim, the operator does not control if, when, or how much arrives, and the pushes can stop as silently as they started. The 5% tax total has not changed.

9 · At review the token was ~14 hours old and every visible trade was manufactured. All 228 observed swaps were made by tax-exempt insiders at 0% tax (zero organic trading), so the token's apparent activity says nothing about how it behaves for real taxed users. Thin pool, no operating history.

The vault's own fee is 0.5% of notional per trade, hard-coded and immutable, its chill-zone width is 17%, and every deposit restarts the 24h cooldown for that wallet's whole position. ⚠️ Never send RADX straight to the vault address: a bare transfer is a donation to all participants, not a deposit, and RADX taxes your donation 5% on the way in. Only deposit() creates a position.

The GEARS vault is deployed and founders-only: a two-wallet founder vault confirmed at block 27,283,417 on August 14, 2026, through a recovery-signed guarded deployment read back on two independent RPCs. It has no deposit function and no public shares — nobody but the two founder wallets can ever put capital to work here or take it out, so nothing on this page is an invitation to deposit. It trades fully taxed at ~$1 per trade as a knowingly-paid volume engine for the token's own ecosystem, not a profit strategy. The findings below were reviewed on July 26, 2026, accepted by the operator and the founder, and re-verified at deployment (last verified 2026-08-14):

1 · The token's tax rate has NO upper limit and the owner can change it silently. Today GEARS charges a measured, exact 5% on every ordinary transfer. But one owner transaction — with no cap, no timelock, and no event emitted — can set the rate to 100% (every transfer confiscated) or even above 100%, which permanently bricks every ordinary transfer, including this vault's sells and your withdrawals. This is the harshest single control surface of any token Tax Haven has ever integrated. In 70 days the owner has never touched the rate; the vault's engine will watch the fee, the fee receiver, and the owner every cycle and halt trading the moment any of them changes — but a change can land faster than any monitor can react.

2 · Two wallets hold two-thirds of the supply. The owner account alone holds ~54% of GEARS and one other wallet ~14%. Either could sell into the thin pool at any time.

3 · 100% of the pool's liquidity is removable, 99.4% of it by one unverified contract. No LP is burned or locked. Nearly all of it sits in an unverified staking contract owned by the token's deployer; its behavior cannot be read from source, and one transaction could effectively drain the venue.

4 · The token sells its own tax inventory inside sells, with zero minimum output. When the token contract has accumulated ≥100,000 GEARS, the next sell into the pool nests a fixed-size, no-slippage-protection dump of that inventory inside the seller's transaction — sandwich-able by anyone watching the mempool, and it adds gas to the vault's sell leg. Milder than RADX's version (fixed size, one hop), and currently disarmed, but it re-arms as fees accumulate. The owner can also repoint the fee receiver at an address that rejects payment, which would make every fee-processing sell revert.

5 · The economics don't clear the friction — that's WHY this is a test. A full round trip costs ~10.79% (5% token tax each way plus four AMM hops via WPLS; there is no usable GEARS/DAI pool), the venue holds only a few thousand dollars a side, and volume is roughly $130/day. At any chill-zone width that clears that friction, the modeled outcome is a fill every few months at best. The one change that would transform this — the owner tax-whitelisting the vault (friction would drop to ~1.2%) — has not been granted, and if ever granted is revocable in a single transaction.

6 · GEARS taxes the founders' own funding and withdrawals: 5% each way for the non-whitelisted wallet. The token charges its tax on every ordinary transfer, including transfers into and out of this vault. The founder MAIN wallet keeps 95% per leg (fund-then-withdraw keeps 90.25%); the founder OWNER wallet is tax-whitelisted on its own token and moves GEARS untaxed. Finding 1 means those are today numbers, not promises.

7 · What's clean, for balance. The token's verified single-file source has no proxy, no reflections, no blacklist, no transfer limits, and a hard-capped supply (no mint path; ~2.7% has been voluntarily burned so far). The 5% was measured at exactly 5.000000% across 364 real transfers. GEARS fails on control and liquidity, not on code quality.

The deployed vault's own fee is 0.5% of measured notional per trade, hard-coded and immutable, sweeping to a cold fee wallet no founder withdrawal can touch. The trading key can trade but can never withdraw; the two founder wallets can each withdraw everything, but only ever to themselves. ⚠️ If you are not one of those two wallets, never send anything to this vault address — the contract has no deposit function, a bare transfer creates no claim of any kind, and it cannot be returned.