Holder protections locked
Agent-ready
4
Holder protections no setting can turn off
10k
Randomised attack runs per mode, on every build

Built on

  • Uniswap v4
  • Ethereum
  • Programmable
  • Foundry
  • wagmi + viem

Agents tune the pool. What holders were promised stays locked. Not the agent, not the creator, not us.

The shell

Four rules. Never switched off.

Every FluxPool has two parts. The core is where blocks, playbooks and agents do their work. The shell wraps it and checks four rules on every swap. If the core misbehaves, or fails outright, the shell still holds.

Iridescent glass structure on black

Checked on every swap

Fees stay under the cap
A launch-window cap, then a normal cap. No block, agent or playbook can charge more.
Sells always go through
No setting, block or failure can stop a holder from selling, in any mode.
Liquidity stays locked
The shell holds the launch position. Nobody can pull it early, the creator included.
No new tokens, ever
Fixed supply. Nothing in FluxPool has the right to mint.

Freedom modes

Pick how much freedom the agent gets.

The community chooses the mode. The four rules hold in all three. What changes is how fast new settings and new code can go live.

Guarded
Riskier changes wait out a timelock, so holders see them coming and can leave first. New code also has to pass compile, interface, simulation and source-verification checks.

Good for

Community tokens and anyone launching for the first time.

Fast
Riskier settings apply as soon as they're signed. New code still has to pass every check, but doesn't wait.

Good for

Teams that react to the market daily and have earned holders' trust.

Wild
Settings and code go live immediately. The shell's four rules are the only guardrails, which is why nobody can turn them off.

Good for

Experiments, and agents you're prepared to let off the leash.

A failed check blocks a change for good, and switching to a safer mode also slows changes already in the queue. The exact timing of every change, mode by mode, lives in the documentation.

Iridescent light streaks on a dark surface
  • One action a minute, 48 a day, by default
  • Removing an agent is instant in every mode
  • Agents can't touch the fee split or the mode

Agents

Plug in any agent. Keep the off switch.

An agent is just an address, so a scoped smart account with a session key works fine. Give it one pool and one of three levels. Raising its level always waits out a timelock.

Fire

Runs playbooks you've already approved, within its rate limits.

Propose settings

Suggests new values. They go live on your mode's schedule.

Propose code

Suggests new blocks. They face the same checks as yours.

Connect an agent

How it works

From launch day to a bad day.

Chrome spheres reflecting a spectrum of light01
Launch through Programmable
One nonce-pinned plan deploys the shell, seeds liquidity it keeps locked, and makes the first buy. Every address is known before anything is signed.
A glass prism splitting light into colour02
Stack blocks into lanes
Before-swap and after-swap lanes, up to four blocks each: anti-snipe, adaptive fees, creator fees, or your own. A block that fails is skipped, not fatal.
Candlestick chart on a dark screen03
Write the playbook for a bad day
Up to eight tighten-only actions that fire on a sustained price drop, then expire and restore themselves. A one-block dump can't set one off.

The FluxPool token

Our own pool runs on the same rules.

FluxPool's token launches through the same shell as everyone else's, with its economics written into the contracts rather than a blog post.

5% → 3.2%

Our fee across the launch window, then flat. Programmable's 0.30% sits on top.

80 / 20

Fee split between the team Safe and the protocol.

5–20%

Of fees go to buybacks, scaling up with market cap.

30%

Goes to buybacks during a sustained drop. Everything bought back is burned.

Ticker and launch date to be announced.

Set the rules once. Let the agent work inside them.