Architecture Proposal: Decoupled PQC (ML-DSA) Shield for OP Stack Pre-Batcher

[SpirosDR1]

Hello OP Labs core contributors,

As the "Harvest Now, Decrypt Later" quantum vector accelerates, we noticed a critical gap in L2 architectures: processing Post-Quantum Cryptography (like NIST ML-DSA) on-chain will inevitably bottleneck the sequencer and disrupt current OP Stack flows due to heavy signature sizes.

At Qubex Sentinel, we engineered a completely decoupled middleware shield. Instead of forcing the OP Stack to handle the cryptography, we intercept, verify, and wrap states at the pre-batcher level.

We just ran a heavy Devnet stress test (1.1 Million concurrent ML-DSA operations across 11 EVMs).
The results directly impacting the OP Stack:

• Base baseline execution: 762,183 ns (0.76 ms)
• Optimism baseline execution: 1,058,173 ns (~1.05 ms)

How it integrates with the OP Stack:

1. Zero dApp migration required.
2. Sequencer logic remains untouched and blazing fast.
3. Completely abstracts the PQC verification overhead from the consensus layer.

We are opening this discussion because we want architectural feedback from the core OP team regarding the middleware interceptor logic and memory pool state-handling before full signature commitment.

Our Devnet logs and implementation architecture are open for review here:

https://github.com/SpirosDR1/Qubex-PQC-Benchmarks

Looking forward to your technical scrutiny. How can we best align this decoupled approach with the upcoming OP Stack roadmap?