Aleph: Intent-Based MEV Solver Agent
The context
Urani.trade is an intent-based trading protocol on Solana: users express swap intents that are batched and settled by competing solver agents, an architecture that reduces toxic MEV at the application layer, protecting users from sandwiching and improving execution. I worked on this as MEV Scientist Engineer on the Agents team.
The problem
The protocol needed an in-house solver: an agent able to ingest order batches, match compatible intents peer-to-peer, and route the remainder across fragmented AMM liquidity, fast enough to compete for order flow.
What we did
I co-authored Aleph, Urani’s in-house arbitrage agent. Aleph fetches order batches from the orderbook, extracts the intents, settles compatible orders through peer-to-peer matching, spins a thread per unmatched intent to compute best quotes via AMM arbitrage (Jupiter routing in this release), and packs the solutions back to the protocol.
Deliverables
- Open-source agent template: github.com/urani-trade/solana-mev-agent-py (Apache-2.0).
- Modular codebase: agents, Solana wrappers, order/intent processing, liquidity venues, P2P matching, price oracles (Helius, Pyth, Dexscreener).
- Tooling: deployment CLI, local protocol-mimicking server for end-to-end testing, pytest suite, Poetry packaging.