Prime IR

https://github.com/fractalyze/prime-ir is an MLIR dialect for cryptography. It is used by a few groups in the ethereum ecosystem.
I have a superset of a subset of it in my lean library that started as a port/bridge of our AUCurves library https://github.com/AU-COBRA/AUCurves. This includes a partial port of the jasmin compiler.

Would there be an interest in having (something like) Prime IR in Veir? If so, what would be the best steps towards that?
Note that for cryptography robust compilation is very important. I have a solution for that in lean, similar to what jasmin does.
And somewhat similar to https://blog.trailofbits.com/2025/12/02/introducing-constant-time-support-for-llvm-to-protect-cryptographic-code/
Do you have plans for adopting such guardrails in VeIR?

My design is inspired by the Com way of structuring dialects from lean-mlir. Do you have plans to expose that abstraction in VeIR?

Start by reviewing the linked Prime IR, AUCurves, and Jasmin projects alongside the issue's Lean-library context. The issue asks for a feasibility and design discussion about adopting a cryptography dialect, robust-compilation guardrails, and Com-style abstractions in VeIR; done would be an agreed direction and next steps.

Written by the indexing model from the issue text.