// the find
ProjectOpenSea/seaport
Seaport is a marketplace protocol for safely and efficiently buying and selling NFTs.
Seaport is OpenSea's on-chain marketplace protocol for trading NFTs and other tokens. It models trades as orders with arbitrary offer and consideration items, letting you do simple fixed-price sales, bundle trades, dutch auctions, and complex multi-party matches all through the same interface. It's for developers building NFT marketplaces or aggregators who need a battle-tested, gas-optimized base layer.
Deployed and actively used across 20+ EVM chains with deterministic addresses, so you're building on something that has handled billions in volume rather than a greenfield contract. The conduit system is a smart design — it separates token approval management from the core protocol, so users can grant approval once to a conduit and use any marketplace built on Seaport without re-approving. Gas optimization is taken seriously: the repo has gas snapshot tracking, optimizer run scripts, and requires gas improvement evidence in PRs. The test suite is genuinely thorough — 100% coverage requirement enforced, dual Hardhat/Foundry suites, and a full fuzzing engine (`FuzzEngine.sol`) that generates randomized order combinations rather than just hitting happy paths.
Versions 1.2–1.4 have known limitations with restricted and contract orders, and they're still deployed on most chains — if you're building an aggregator you have to handle the version differences or you'll hit edge cases in production. The architecture is split across four separate repos (seaport-core, seaport-types, seaport-sol, seaport) as git submodules, which means understanding the code requires chasing down four codebases; the README doesn't explain this upfront. The zone and conduit extension points are powerful but under-documented — `ZoneDocumentation.md` exists but the actual interface behavior around `validateOrder` return values and when zones get called is non-obvious without reading the source. Last commit activity has slowed down considerably compared to the 2022–2023 peak, so don't expect rapid iteration or quick issue responses.