In a November 2 UTC research post, Harry Jeon presented an OP Stack proof-of-concept that separates transaction authentication evidence from the data carried by individual transactions. The Ethereum Research post describes a devnet implementation and links a public code fork.
The prototype adapts EIP-8288, a proposal for aggregating signatures and proofs. Its upstream specification remains Draft. Jeon's implementation changes where the aggregate travels: it covers several L2 blocks and is submitted to Ethereum through a new OP Stack batch type, rather than accompanying each block in its header.
In the described transaction flow, a user sends a transaction and its supporting proofs to the sequencer through RPC. The transaction declares the statements it depends on. A separate process, op-aggregator, combines the corresponding evidence across blocks, and the batcher posts the resulting aggregate to L1.
The important implementation boundary comes later, during derivation, when a node reconstructs the L2 chain from L1 data. A new dependency gate looks ahead for the aggregate covering a block. The post describes retaining the batch when its dependencies are present and replacing it with a deposit-only block when the required aggregate or statements are missing. These are the prototype's described derivation rules.
The adaptation also differs from the L1 draft's transaction interface. EIP-8288 builds on frame transactions, which separate validation, payment approval and user operations into calls. The prototype instead uses an account-abstraction transaction format and a single proof scheme.
For client developers, the design therefore spans the sequencer, aggregator, batcher and derivation pipeline. Jeon presents the chain as an optimistic rollup with fault proofs; proving state transitions and moving to a fully zero-knowledge rollup are listed as future work.