A memecoin that trades.

$CATGENT launches on letscash.fun, on Robinhood Chain. Every trade of the token generates creator fees in ETH. Instead of those fees going to a dev wallet, they fund an autonomous agent that hunts newly-launched tokens on the same launchpad — and converts what it earns into $CASHCAT, the original meme of Robinhood Chain.

Every decision it makes is published live, including the ones where it decides not to act.

The loop

$CATGENT trades on letscash.fun
        │
        ▼   creator fees accrue in ETH
   the agent claims them
        │
        ▼   working capital
   snipes new launches on letscash.fun
        │
        ▼   profit
   converts to $CASHCAT
        │
        ▼   swept to a cold wallet it cannot withdraw from
   goal: 10,000,000 CASHCAT — 1% of supply

ETH is the tool. CASHCAT is the objective, and it is never sold — so the scoreboard is a token balance that only goes up. There is no way to "win" by sitting in cash, and a flat week reads as slow progress rather than as a smaller number.

The mission

10,000,000 CASHCAT — exactly 1% of the 1,000,000,000 supply. Split into ten milestones so the bar actually moves:

10k25k50k100k250k 500k1M2.5M5M10M

Milestones latch — once reached they stay reached, so progress never appears to reverse. The dashboard tracks the current rung rather than the 10M endpoint, because progress toward 10M would sit near zero for months and tell you nothing.

Holder rewards

Every milestone the agent reaches, 5% of it goes back to $CATGENT holders — ten wallets, drawn at random, paid in CASHCAT. Holding through the whole run is the point.

Eligibility: hold at least 50,000 $CATGENT. An even draw across every wallet would reward splitting one bag into fifty, so there is a floor — low enough that ordinary holders qualify, high enough that farming the draw costs real money in both tokens and gas.

Milestone	Shared	Per holder
10,000	500	50
100,000	5,000	500
1,000,000	50,000	5,000
10,000,000	500,000	50,000

The draw is verifiable. Winners are picked by a deterministic function seeded with a block hash chosen after the milestone is reached. The seed, the block number and the eligible-holder count are all published on the dashboard — so anyone can re-run the draw and confirm the same wallets come out. A draw seeded with a random number nobody can check would be indistinguishable from picking friends.

Contracts, the trading wallet and the vault are never eligible. Funding is manual on purpose: the agent marks what it owes and waits for the amount to arrive from the cold wallet — it holds no key for the vault, so it cannot start a payout on its own.

How it picks trades

There is no indexer for these pools, so both core signals are derived from chain data. The launchpad's hook charges a known percentage of every swap and emits an event — so gross fee ÷ fee rate is the traded volume, and one log query returns it for every pool on the launchpad at once. Hundreds of tokens get ranked from a single request; only the top handful cost a price quote.

Each candidate gets a 0–100 score from six weighted factors:

Factor	Weight	What it reads
Volume acceleration	24%	the last minute against its own 15-minute baseline
5m momentum	20%	price change — on an inverted U, not a ramp
Absolute volume	16%	real ETH through the pool in 5 minutes
Pool depth	14%	how much it can absorb
Fee drag	14%	inverted — cheap pools score, expensive ones don't
Entry timing	12%	prefers a pullback to buying the exact top

Why momentum peaks and then falls. A monotonic ramp gives its best mark to the most extreme move on the board, which is backwards on pools this thin. Caught in testing: a token printing +118% in five minutes took full momentum marks, scored 56, and lost 31% in twenty-seven seconds — while the timing factor was simultaneously reporting "chasing the top". Anything past +45% is now discounted as a blow-off. That same trade scores ~40 today and is skipped.

Why fee drag carries real weight. The launchpad publishes 1%, 3%, 5% and 10% fee tiers. The fee is charged on both legs, and slippage is given up twice — so a round trip costs about 5% on a 1% pool and 23% on a 10% pool, measured live, before the price has moved at all. Gas on this chain is ~0.0000026 ETH and is irrelevant by comparison. The pool fee is the whole cost, and a strategy that ignores the tier bleeds out on structurally unwinnable trades.

Hard vetoes

Kept separate from the score, because a veto is a fact about whether a trade is possible and a score is an opinion about whether it is attractive. Mixing them lets a strong opinion outvote a fact. A candidate is refused outright if:

it is outside the launch-age window
pool depth is below the floor
the intended size would move the price past the impact cap
there is not enough price history to judge direction
the profit target does not clear the pool's round-trip cost with margin
a simulated round trip returns materially less than the pool's own fee explains — the book is too thin to take the size and give it back

The exit ladder

Most new launches go to zero. The arithmetic only works if the stake comes back early and the survivors are still held when they run.

1 · Stake at risk Stop-loss live. A time stop closes anything that never clears its round-trip cost.

2 · Take initials at 2× The fraction sold is solved, not assumed: (1+buffer) ÷ (multiple × (1−fee)). On a 1% pool that is 53%, not the 50% folklore assumes — and the difference is exactly the sell fee, which is why "sell half at 2×" quietly leaves you short.

3 · Stop switches off The stake is back; the remainder is house money. The stop and trailing stop are deliberately disabled — that is what lets a moonbag survive the drawdowns that would shake a normal position out.

4 · Scale out at 3×, 5×, 10× Each tier sells a share of the original position, never of what remains.

5 · Moonbag 20% is never sold on price, enforced as a hard floor so the tiers can never eat it. Only a collapse in flow, or an exit that stops working, closes it.

Risk controls

Impact-bounded sizing — not a fixed clip. Measured live, 0.25 ETH moved one pool 4% and another 18%, so the binding constraint is the impact the trade itself causes.
Position cap — a maximum number of concurrent positions, and a ceiling per position as a share of equity.
Daily loss limit — halts new entries for the rest of the UTC day.
Gas reserve — a floor the agent never spends below.
Working-capital floor — CASHCAT conversion only ever takes the surplus, because trading is what earns the next CASHCAT.
Kill switch — on the dashboard, and as a file that works even if the dashboard is unresponsive.

Where the CASHCAT goes

Acquired CASHCAT is swept out of the trading wallet to a cold wallet the agent has the address of but no key for. It can send; it can never withdraw.

That asymmetry is the point: a total compromise of the trading wallet costs the ETH sitting in it, not the mission. The scoreboard counts both wallets, so a sweep neither adds nor removes progress — it just moves what is already owned somewhere safer.

How the numbers are kept honest

Fee claims are capital, not profit. A claim raises equity with no trade behind it, so it is tracked separately and subtracted out of every performance figure.
Gas is an operating cost, not a trade cost. Folding it into a trade's P&L would let a busy block turn a winner into a loser.
Positions are marked at the exit, not the quote. Each is valued at the ETH a full sell of exactly that size would return — marking at the quoted price would systematically overstate equity in pools this thin.
Deposits and withdrawals are detected and excluded. The wallet is reconciled against the chain every tick; anything moved by hand is booked as capital, never as strategy P&L.
Every value is stored as an exact integer. Wei does not fit in a floating-point number, and a silently-rounded balance is the one bug a ledger must not have.

Honest limitations

The 10,000,000 CASHCAT goal is a target, not a forecast. Nothing here makes it likely.
This is a high-variance strategy on thin, brand-new pools. Most launches go to zero. The risk limits are defaults, not advice.
Pool depth is an estimate derived from measured price impact, not a reserve read directly.
The agent can see price, volume, liquidity and fees. It cannot see intent — it does not know whether a dev is about to sell.
Simulated results are never published here. Performance figures stay empty until the agent trades with real funds.

None of this is financial advice. $CATGENT is a memecoin.