MOIRAIS: A Multi-Domain Scientific Computing Toolkit for Observational Inference, with Sociolegal, Signal-Processing, Cryptographic, and Spatial-Statistics Modules

I describe MOIRAIS (Methods for Observational Inference and Robust Analysis of Interventions in Scientific Experimentation), a dual-language (Python and R) open-source scientific-computing toolkit. MOIRAIS is multi-domain by construction: a unified package surface supports general-purpose causal inference and survey-weighted estimation, signal processing and spectral analysis (with applications to forensic audio and biomedical signals), classical and modern cryptographic primitives, spatial statistics, statistical-physics models of crime, classical-test-theory and item-response-theory psychometrics, automated scrapers for Canadian carceral and police-oversight corpora, and a flagship implementation of the MRM (McNamara-Ruhela-Medina) framework for sociolegal data analysis. The MRM framework's mathematical foundations are developed in detail in the companion paper at 10.5281/zenodo.20096075. The package ships numerous built-in datasets in a portable SQLite layer, a 36,476-file individual-function namespace, a 10-screen Textual terminal user interface, a multi-provider local-LLM chain with a vendored TurboQuant KV-cache compression, and a pure-Python emissions tracker covering 213 countries. The current Python test suite contains an extensive baseline test suite plus 55 tests covering the federal/provincial Mandela cross-jurisdiction modules, all passing under continuous integration. The package is GPL-2.0-only (Linus copyleft, deliberately chosen over GPL-3.0 for compatibility with the broader Linux-kernel-style ecosystem) and is released via PyPI, CRAN, and r-universe simultaneously. The R package mirrors the Python package function-for-function.