Getting Started

Protocol Overview

x0x1 is a Calldata-as-a-Service platform for DeFi yield. AI agents verify APY from smart contracts, simulate on Anvil forks, and present state diffs for self-custody signing.

Strategy Architecture Key numbers Trust model Transparency

x0x1 is a Calldata-as-a-Service platform for DeFi yield. AI agents scan 29 protocols across 15 chains and verify every APY directly from smart contract state — not cached APIs, not aggregator feeds. They build transaction calldata, simulate on Anvil forks, and present a full state diff before you sign anything.

You keep your keys. We produce calldata. No custody. No approvals. No access to your wallet. x0x1 outputs unsigned transaction bytes — you verify, you sign, you broadcast. The agent never touches private keys.

SDK Adapters

Chains

Exploit DB

TVL Floor

Strategy

Aggregator APYs are unreliable. Morpho Blue's raw apy field shows 2980% — gross APY before curator fees. The correct net_apy is 53%. That's a 56x error. x0x1 adapters read contracts directly.

Architecture

Three independent, composable pipelines:

Scan

Yield Engine discovers and verifies opportunities across 15 chains. Every APY is verified from smart contract state.

Build

Calldata Pipeline assembles transaction calldata, simulates on Anvil fork, produces full state diff for review.

Sign

Safe Signing sends unsigned TX through air-gapped bridge to isolated signing device. Keys never touch the internet.

Key numbers

Metric	Value	Details
Protocol adapters	29	Standalone Python modules, stdlib only
Chain coverage	15	Ethereum, Arbitrum, Base, Optimism, Polygon, Solana, HyperEVM, and more
Exploit database	523+	Historical DeFi hacks cross-referenced with every position (details)
TVL minimum	$1M	No positions below $1M TVL pass the risk filter
APY verification	Direct RPC	archive RPC nodes, JSON-RPC batch calls
Simulation	Anvil fork	Full state diff before every transaction

Trust model

"Agent as unreliable center" — the AI orchestrates, deterministic code enforces:

Hard rules in code

TVL floors, exploit checks, gas estimation, and approval scoping are enforced programmatically. The agent cannot bypass them.

Halt over guess

Uncertainty = hard stop with a reason code. No fallback heuristics, no defaults.

Intent staging

Natural language intent → typed proposal → simulation → user approval → signing. Each step is auditable.

Determinism ratchet

Capabilities migrate unidirectionally from agent to deterministic code. The system gets more reliable over time.

The agent never signs transactions. It cannot access private keys. The signing device is physically separate, connected only through a unidirectional bridge.

Calldata-as-a-Service. x0x1 is not a custodian. It does not hold funds, manage wallets, or broadcast transactions on your behalf. It produces verified calldata — you sign on your own device. Every transaction goes through the full pipeline described here. No shortcuts. No exceptions.

Transparency

x0x1 is pre-audit and not yet open-source. An independent security audit is planned for Q3 2026, and SDK adapters will be published on GitHub after the audit completes. In the meantime, beta users receive full source access to all adapter modules for independent review.

The signing device runs on your hardware — no hosted service touches your funds.

RPC down? Pipeline halts with HALT_RPC_UNAVAILABLE. No fallback.
Wrong recipe? Reverts on Anvil fork, not mainnet. Self-healing loop patches it.
Exit a position? Same pipeline. "Withdraw from Euler" → calldata → simulate → sign.
Bridge fails mid-signing? Unsigned TX stays in relay buffer. Retry later. No partial state.

Stop trusting aggregator feeds. Verify yield from smart contracts.

Explore SDK Adapters → Signing Architecture → or request beta access