SODAX SDK V2 Brings Smoother Cross-network Execution with Agentic Support
Version 2 of the SODAX SDK closes a gap for cross-network applications who need a scalable solution for liquidity access and defi execution across 18+ networks. Now offered with built in support for agentic workflows for initial launch, and ongoing trade execution.
SODAX Team
•
SODAX SDK v2 is now live, existing partners should begin migrating their SODAX implementations to the new system.
The SDK is now structured to be readable for agents (Claude, Cursor etc), enabling agile integration and ongoing agentic trade.
On v2 of the SDK, partners can look forward to incoming support for an additional 55 assets across our integrated networks.
Cross-network DeFi has had a software problem for as long as it's existed.
The protocols themselves work. Solvers route trades. Lending markets hold collateral. The on-chain plumbing has matured to the point that anyone with a wallet can move value between blockchain networks in an afternoon. But all this provided they're willing to read the documentation, hold the right wrapped assets, and accept the risk that each step in the chain is operated by a different team with different security assumptions.
This software problem is what happens when a builder tries to wrap the diverse experiences of “web3” into a more abstract product their users can actually use. If doing it without an execution brain at the center, focusing on achieving the users intent and working within constraints.
Most cross-network tooling was built under the assumption that the application would handle routing and know which network an asset was on, which path was cheapest, which path might fail, how to handle a half-completed execution. For a single-corridor proof of concept, that's fine. For a real cross-network product, it's a return to battle for each network expansion.
This is the gap that the SODAX SDK was built to close. V2 closes it.
What V2 changes
Stability
V1 grew naturally. New networks got added, New product paths (Lending, Borrowing, Staking) layered on top of cross-network swaps. Each addition came with its own assumptions about state, error handling, and dependency order.
V2 is a refactor in the strict sense: the codebase was taken apart and rebuilt around the integration paths partners actually use, with comprehensive bug fixes, performance optimizations and a cleaner internal architecture. Less code in the SDK means fewer places for the SDK to fail in production.
Dev ex
Integrating V1 was easy, but getting that first cross-network exchange flowing should be better. The SDK was intuitive once you'd internalized it, but the on-ramp has now been improved.
V2 was rewritten with a single rule: a developer should be able to read the entry points and understand what's available to them without reading the docs first. Method names describe what the methods do. Types are explicit. The boundaries between an intent (what the user wants) and an execution (how the system fulfills it) are visible at the SDK surface, not buried in the implementation. Integration paths for wallets, DEXs and routing protocols, lending and money markets, yield strategy and vault builders, are now substantially shorter.
Agent ready
This is the piece that changes who can use the SDK at all.
V2 is structured so that LLMs (see Claude, Cursor, and the rest) can read the SDK, understand its surface area, and produce working integrations into front-ends or agent-driven workflows. The documentation is now written explicitly for agents to be able to consume. Examples are present in the right places to seed the model's understanding without requiring a separate documentation lookup.
The reason this matters is that the population of teams who can ship a cross-network integration is changing. The number of developers who can hand-write one is small. The number of teams who can ask Claude Code or Cursor to wire one up is much larger. And we forsee that the number of agents that will themselves operate as users of the protocol (running cross-network exchanges, managing money market positions, settling between networks) is going to grow faster than the number of human users in the coming months.
We’re betting early that DeFi's next wave of users wouldn't be people. SDK V2 is the integration layer for that bet.
What's live for partners today
All execution paths refactored and tested: cross-network exchange, money market supply and borrow, settlement.
Demo applications migrated to V2 and validated end-to-end.
Type-explicit, LLM-readable structure across the SDK.
Documentation aligned.
What's coming next
V2 ships today and all existing partners should begin migrating their SODAX implementations to the new system. The next iteration of work is on widening the surface area, with readers of our monthly tech update already aware that Vaults and additional assets are on the way. With v2 live, a further 55 leading assets are now in line for support through the SDK and will be shared in the coming days
If you're a builder who wants to integrate, or an agent operator looking to build on top of the SODAX execution system, the integration starts at sodax.com.
You can also catch up on technical updates to the SODAX execution system with our monthly tech updates and their matching Youtube podcasts.
definewsannouncements