Retiring PQClean

[Thom Wiggers]

Hello all,

We have been maintaining the PQClean project since 2019 and over the course of the (initial) NIST PQC competition I feel that it has been very successful in making the reference implementations more accessible for experimentation with integration of PQC primitives into protocols, as well as supporting many research projects who used PQClean implementations as a basis for future work.

We were also able to publish a report on what we learnt from PQClean and how future standardization projects could, with suggestions ranging from small tweaks (enabling -Wall in the examples) to considering other implementation languages than C (Python, Sage, or perhaps Rust), avoid, e.g., safety bugs while simultaneously improving the usefulness of reference implementations. You can find this on IACR ePrint at https://eprint.iacr.org/2022/337.

Unfortunately, PQClean has never really been able to capture the new submissions to the NIST signatures on-ramp, or integrated any of the (inter)national PQC standardization efforts. Most of the implementations in PQClean are (ML-KEM, SLH-DSA), or will be soon (FN-DSA, HQC) be, out of date and superseded by “official” reference implementations and/or high-quality implementations from projects such as the PQ Code Package.

The current state is the result of us longer having the time to maintain PQClean. We have now concluded that it is time to retire PQClean. We are planning to mark the GitHub repository as “Archived” (i.e., read-only) in July 2026. Libraries packaging PQClean source code should migrate to other, maintained, sources ASAP. There are some pointers in our README.

To discuss this, refer to the announcement in https://github.com/PQClean/PQClean/issues/604. 

I would like to thank everybody who contributed to PQClean, in particular co-maintainers Matthias Kannwischer and Douglas Stebila, and our co-“founders” Peter Schwabe and Joost Rijneveld. I am also grateful to all contributors, who can be found at https://github.com/PQClean/PQClean/graphs/contributors, as well as everybody who reported issues and discussed suggestions.

Regards,

Thom Wiggers