ElytroEthereum wallet for
you & your agents
CLI-first. Agent-native. Self-custodial.
Interactive Elytro CLI and agent chat demonstration — CLI on the left, chat on the right — illustrative only
Who controls your AI on Openclaw?
Human-first wallets not suited for agents
They assume a person clicks every button — not agents.
Custodian platforms = centralised control
Giving keys to platforms hands control away from you.
Agents need enforced boundaries
Without on-chain policies, agents operate without boundaries
A smart wallet for humans & AI agents.
Transactions execute under on-chain policies you define.
Built for agents. Controlled by you.
Beta- Works with any AI agents e.g. OpenClaw and Claw
- Your wallet lives in your favourite chatbot
- Create wallets & check balances
- Simulate, send & batch transactions
- On-chain 2FA email authentication
- Enforce your own spending limits
Building your first agent-native wallet
Your AI agent. Your rules. Enforced on-chain.
On-chain policy with spend limits
Define and enforce what agents may (and may not) do.
True self-custody — fully on-chain
Your keys and account live on-chain — safe even if we go offline.
Social recovery & 2FA protection
Multiple, secure recovery paths — no seed-phrase reliance.
One of a kind
Agent-native. Self-custodial. On-chain enforced.
| Traditional wallets | Agent wallets (custodial) | Elytro | |
|---|---|---|---|
| Self-custodial | |||
| Agent-compatible | |||
| Policies enforced on-chain | |||
| Works if provider disappears | |||
| On-chain 2FA & spending limits |
Towards the next-gen Ethereum wallet
The future will not be manual. But it must remain sovereign
Native smart-contract accounts
EIP-4337 + EIP-1271 — real on-chain accounts.
Native agent runtime engine
Account-abstraction + on-chain policy for safe agents (EIP-8004, proposed).
Ethereum-first product roadmap
Built for Ethereum's future; shipping on mainnet first.
On-Chain Agent Operations
Elytro lets AI agents safely execute Ethereum transactions through a self-custodial smart-contract wallet with on-chain policy controls.
Best with Claude Opus 4.5+, GPT-5.3-Codex+, or Gemini 3.1 Pro. As of Mar 2026.
Your questions answered
→More FAQsElytro is the next-gen Ethereum wallet built for the AI era — Ethereum-only, CLI-first, agent-native, self-custodial. One smart account, operated by humans, agents, or both.
Install globally with npm: npm install -g @elytro/cli (requires Node.js ≥ 24). For OpenClaw, use Copy OpenClaw setup prompt on the homepage or the same block under Documentation — it loads the Elytro wallet skill from GitHub and walks you through CLI install, wallet init, a secure OP Sepolia account (email 2FA and daily spending limit), and on-chain deployment.
The Elytro CLI provides a deterministic execution surface that AI agents can use to interact with your wallet. Spending limits, policies, and approvals are enforced on-chain, so agents operate within boundaries you define — while you retain ultimate authority.
Elytro CLI works best with higher-capability models that handle multi-step tool use and on-chain reasoning. As of March 2026, we recommend Claude Opus 4.6, Claude Opus 4.5, GPT-5.4, GPT-5.3-Codex, or Gemini 3.1 Pro.
You can configure spending limits (per-transaction and daily caps), allowlisted addresses, permitted transaction types, and approval requirements. All policies are enforced on-chain by your smart account, not by Elytro's servers.
We believe in GUIs — but we think the GUI for the AI era needs to be reinvented. The Chrome extension remains available for existing users, but we're rethinking the interface layer for a world where humans and agents share execution authority. The new GUI is in active development.
Yes. Your account is 100% on-chain and truly self-custodial. You can run a local client or interact with your smart contract directly — even if we disappear. True self-custody means you're never dependent on us.
Vitalik Buterin