For autonomous agents & intent solvers
Cross-chain agent execution
Give an autonomous agent one all-or-nothing primitive for acting across many rollups at once — submit a cross-chain intent, settle a deterministic outcome, never a half-executed position.
A 2026-era agent managing capital across Base, Arbitrum, and Optimism has to act on several chains inside one decision. Wiring that from independent bridge transactions means every leg can land or revert on its own, and the agent is left reconciling a broken, partially-executed state — exactly the failure mode autonomous execution cannot tolerate.
- 01
The agent (or an intent solver acting for it) expresses the desired outcome as a single atomic swap group bound by one shared swap_group_id and a deadline.
- 02
The Vyper contracts enforce all-or-nothing settlement: every leg resolves within the window, or every leg refunds — so the agent settles a deterministic outcome, not a set of independent bets.
- 03
Commit-reveal hides the intent on every chain until block ordering is fixed, so an agent's strategy is not front-run the moment it is legible in a mempool.
- 04
The relayer submits the legs but cannot authorise resolution — the agent never has to trust an off-chain solver or bridge operator to be honest for the guarantee to hold.
Build this on Tesseract.
The protocol is open source and self-hosted. Start from the quickstart, or read how the atomic-swap lifecycle works under the hood.