Dumb Wallet
Autonomous Wallet using AI and ENVIO Hyperindexing.
视频
技术栈
描述
Dumb Wallet
Autonomous, verifiable DCA and portfolio automation on Monad. You set the rules once; the agent executes within strict, auditable limits.
What you get
Automated buys, rebalancing, and Magma restaking (gMON) using real-time on-chain signals and whale activity.
Best‑execution routing across multiple DEXs to reduce slippage and gas.
Transparent AI plans with rationale before execution (you always know why).
How it works
You choose a trading style (Conservative, Balanced, Aggressive, Contrarian).
The agent analyzes portfolio balance, market momentum, sentiment, and whale flows via Envio HyperIndex, proposes actions, then executes within permissions you approve.
Delegations = pre‑approved rules about what the agent can do, for how long, and with what value limits. Account Abstraction (AA) = smart account logic that enables sponsored (gasless) execution and policy enforcement.
Trust and safety (verifiable by design)
Fine‑grained permissions (MetaMask Delegation Toolkit, function‑level selectors).
Time‑locked delegations and value caps (TimestampEnforcer, ValueLteEnforcer).
Panic mode: instant revoke, convert, and withdraw.
Full audit trail of AI decisions with confidence scoring.
Under the hood
ZeroDev AA for sponsored execution and policy‑driven transactions.
Envio HyperIndex for real‑time metrics + whale registry.
Multi‑protocol routing and analytics: Uniswap V2, Kuru OrderBook, Ambient CrocSwap.
10+ tokens: MON, WBTC, USDC, CHOG, DAK, BEAN, etc.
Magma restaking pipeline: automated stake/unstake MON → gMON.
AI: Dumb - Centralized with DeFi‑tuned prompts ... Smart - Designed for custom models, inference swarm‑ready architecture (Fortytwo/OpenGradient).
Execution logic: price triggers, trailing DCA stops, EMA gates; emergency revocation.
Quick start
Create ans sign your MetaMask Smart Account
Fund your bag with MON
Toggle Magma restaking
Pick a trading style
Let it run (you can Panic‑revoke anytime)
Built for Monad testnet with native integrations. The core is a delegation‑based execution model: you pre‑define exactly what’s allowed, the agent proves compliance, and every action is verifiable.
If you want to run yourself please check Github repo instructions.