Most analyses treat their inputs as exact. When the inputs are noisy, the honest answer is often weaker, and occasionally stronger, than the one first reported.
What it does, and how
You supply data with an uncertainty for each point. It returns a corrected fit with convergence checks, a description of what is left unexplained, and a report that separates what was tested from what was assumed. PCR-2 is one of the methods it can use.
Where it is up to
The software is built and tested, which is engineering verification and not scientific validation. Results produced before the September 2026 correction are treated as historical until rerun, and the reruns completed so far are mostly null or bounded. It is private, with no public release.
Where you come in
If a conclusion of yours rests on measures you know are noisy, I can talk through whether this kind of analysis would change it.