feat: deploy UniversalRouter 2.2.0 to mainnet, sepolia and ink

[dianakocsis]

Deploys UniversalRouter 2.2.0 to mainnet, sepolia and ink, and records the results.

Deployments

Chain	Address	Verified
Ethereum	0xab863E752Bf67D8DCDD929EaAe9Be9dc83Fb3BbB	Etherscan
Sepolia	0x5093f1CDED83d99FfEd6602dA6260672ae16787c	Etherscan
Ink	0x8BbCAEb326AD590521a921C9e457Ee09F336B48c	Sourcify

All three were deployed from universal-router tag 2.2.0 (64027f3) via Deploy-all, using CREATE2 with a zero salt through the deterministic deployer, matching how the previous 2.2.0 routers were deployed.

Each address was computed from the initcode before broadcasting and matched what the deploy produced. All three carry a 23,233-byte runtime, which matches the snapshot committed in Uniswap/universal-router#514. Every constructor parameter was taken from the previous 2.2.0 deployment transaction on that chain, checked to have code onchain, and cross-checked — v4PositionManager.poolManager() resolves to the v4PoolManager argument on all three, and the v2 pair init code hash was confirmed by predicting a real pool address from it.

Submodule and briefcase

src/pkgs/universal-router moves from fb25ff0 to 64027f3. That brings the Dispatcher change from universal-router#499 and moves v4-periphery from 07336f21 to a7af5b34.

The deployer initcode is regenerated to match: 24,102 to 25,193 bytes. Only the hex literal is updated — insert_initcode.py rewrites the surrounding formatting in a way that fails forge fmt, so the rest of the file, and the 30 other deployers it touched without changing their initcode, are left alone.

Build fix

src/pkgs/liquidity-launcher has no entries in the root remappings.txt, only in its own, so forge build cannot resolve its imports. It has been unbuildable since it was added in November 2025.

That was harmless until e59f730 rewrote the flatten step to run in parallel and abort on failure. Since then every create_briefcase.sh run has died on that package, which is why the committed initcode had not moved since 2026-05-27 and the Update Briefcase workflow has been failing.

Wiring it in properly is not a remapping away: process_briefcase_files.py resolves its @uniswap/v4-periphery imports against the top-level package rather than its own lib, and its nested dependencies pull in further unmapped libraries. The package has also never been part of the briefcase. Skipping it in both places restores the pre-e59f730 behaviour; the briefcase skip list is overridable via BRIEFCASE_SKIP_PKGS.

Note for anyone repeating this

forge 1.8.1 silently degrades the CREATE2 in DeployerHelper into a plain CREATE on ink, producing a nonce-derived address with no error. forge 1.5.0, which this repo's CI pins, handles it correctly. Compilation is unaffected — the deployed bytecode on mainnet and sepolia is byte-for-byte identical to a 1.5.0 build once the 49 immutable slots are masked out.

🤖 Generated with Claude Code