Invariant Labs LLC

Deterministic verification infrastructure for software supply chains.

About

Invariant Labs builds foundational infrastructure where correctness is the product. We develop deterministic data processing libraries, standards-grade conformance tooling, and engineering practices that produce evidence of correctness.

Our work operates at the lowest layers of the software supply chain: IEEE 754 floating-point formatting, JSON canonicalization (RFC 8785), and deterministic build verification. The stack invariant is simple: same inputs and same tool identities produce byte-identical governed outputs.

Work

json-canon

Go / Apache 2.0 / github.com/lattice-substrate/json-canon
Pure Go implementation of RFC 8785 (JSON Canonicalization Scheme). ECMA-262 Number::toString, strict RFC 8259 parser, zero dependencies. 286,362 cross-engine oracle test vectors.

Engineering Articles

lattice-substrate.github.io
Technical writing on deterministic data processing, IEEE 754 digit generation algorithms, strict JSON parsing, infrastructure architecture decisions, and evidence-based release engineering.

JCS Conformance Research

Zenodo / DOI: 10.5281/zenodo.18890836
Cross-language conformance evaluation of RFC 8785 implementations. 4×2 design across four algorithms and two languages (Go, Rust), eight implementations. Identified subnormal formatting failures in two reference implementations.

Lattice Substrate

GitHub / github.com/lattice-substrate
Organization home for deterministic infrastructure primitives. Open-source libraries and conformance tooling.

Research

Fixed-Width vs. Multiprecision Digit Generation for JSON Canonicalization. Under peer review, Software: Practice and Experience (Wiley).

Conformance dataset: doi.org/10.5281/zenodo.18890836

Upcoming

IIW 42 (Internet Identity Workshop), April 28 to 30, 2026, Mountain View, CA. Presenting on JCS number formatting conformance.

Contact

mark@provemark.io
Mark Lenhardt / Founder, Principal Engineer