AgentE plugs into your game server, watches your economy every tick, detects imbalances, simulates fixes, and applies them — or recommends them for your approval.
Balances, transactions, inventories, prices — whatever your game tracks. AgentE connects to your database and reads it every tick.
Up to 60 economic principles run against your live data. 5 free, 60 with Pro. Inflation creeping up? Wealth concentrating? A sink drying out? It catches it and tells you why.
It simulates the fix (100+ Monte Carlo runs), proves it won't make things worse, then either applies it automatically or waits for your approval.
LIVE FEED — AgentE detecting and resolving violations in real time
HOW IT WORKS
Connects to your game database and reads balances, trades, inventories, and prices every tick.
Checks your active principles against live data. 5 in Community, 60 in Pro. Catches violations and ranks them by severity.
Before changing anything, it runs 100+ Monte Carlo simulations to prove the fix works. No guessing.
Picks which lever to adjust and by how much. Respects locked parameters, bounds, and cooldowns.
Applies the change to your game. Watches every tick after and rolls back anything that backfires.
THE RULES IT FOLLOWS
AgentE doesn't use AI to guess what's wrong. It checks your economy against hard rules — 5 in the free Community tier, 60 with Pro — each one designed to catch a specific failure mode before it spirals.
INTEGRATION
import { AgentE } from '@agent-e/core';
const agentE = new AgentE({ adapter });
agentE.connect(adapter).start();
// That's it.const agentE = new AgentE({
adapter,
mode: 'advisor', // 'autonomous' on Pro
gracePeriod: 50,
cooldownTicks: 15,
onDecision: (d) => log(d.reasoning),
});agentE.lock('your_fee_param');
agentE.constrain('your_cost_param', {
min: 0.5, max: 2.0
});
agentE.on('beforeAction', (plan) => {
if (plan.targetValue > 2.0)
return false; // veto
});YOUR GAME, YOUR RULES
Every game has rules unique to its design. Write your own principles and AgentE enforces them alongside the built-in ones — same pipeline, same simulation proof.
agentE.addPrinciple({
id: 'MY_RULE_1',
name: 'Support roles must always exist',
check: (metrics) => {
const share =
metrics.roleShares['support'] ?? 0;
if (share < 0.05)
return {
violated: true,
severity: 8,
evidence: { share },
};
return { violated: false };
},
});Not ready for autonomous? AgentE tells you what it would do — with simulation proof — and waits for your sign-off.
const agentE = new AgentE({
adapter,
mode: 'advisor' // ← never acts unilaterally
});
agentE.on('decision', (entry) => {
console.log(entry.reasoning);
// entry.plan.simulationResult has full proof
if (youApprove(entry))
agentE.apply(entry.plan);
});Real-time charts, decision feed, alerts, persona distribution — all at localhost:3100. No extra setup.
import { startServer } from '@agent-e/server';
const server = await startServer({ port: 3100 });
// Open http://localhost:3100 — dashboard is live
// Health charts, decision terminal, alerts,
// persona distribution, violation history.
// WebSocket real-time. HTTP polling fallback.
// In advisor mode: approve/reject from the UI
// curl -X POST /approve -d '{"decisionId":"..."}'AgentE explains its decisions in plain English. Optional — bring your own LLM provider. Zero cost to AgentE.
const agentE = new AgentE({
adapter,
llm: {
provider: myLLMProvider, // any backend
features: { anomalyInterpretation: true }
}
});
agentE.on('narration', (n) => console.log(n.narration));
agentE.on('explanation', (e) => console.log(e.explanation));
agentE.on('anomaly', (a) => console.log(a.interpretation));When a violation fires, the LLM explains it in plain language. "Inflation is spiking at 12% because faucets are outpacing sinks by 3:1."
Before applying a fix, the LLM explains what will change, what the simulation predicts, and what could go wrong.
Detects statistical anomalies the principles don't cover. Rolling mean + stddev, then asks the LLM to interpret unexplained deviations.
PRICING
No monthly fees. Community is free forever. Pro charges only when AgentE changes a balance in your economy.
ROADMAP
The core economic intelligence engine — 5 Community + 55 Pro principles, 5-stage pipeline, and Monte Carlo simulation. Built for game economies that need autonomous balancing.
Extending AgentE into decentralized finance — autonomous liquidity management, yield optimization, and protocol-level economic guardrails deployed directly on-chain.
A marketplace for economic intelligence — share principles, trade adapters, and license battle-tested configurations across every ecosystem.
Start free with Community. Scale to 60 principles with Pro — pay only when AgentE changes a balance. No monthly fees.