characterization-testing

A Claude Code plugin marketplace for open-source maintainers — README heroes, README audits, and .NET test-coverage backfills.

An agent skill that refactors legacy code without changing its behavior — retained characterization tests, precise names, locality over abstraction.

Multi-mode crystal oscillators tuned for Stratum 2 Rubidium-class clock stability, at a fraction of the cost. Funded by NLnet Foundation.

Claude Code plugin: Generate characterization tests for legacy code before AI modification. Supports Go, Python, TypeScript, JavaScript, Kotlin, Java.

Byte-level proof that a refactor changed nothing: golden-census records a program's stdout, stderr, exit codes and written files with the machine normalized away, and compares them after a change. Built for LLM coding agents; Python 3.11+, no dependencies.

Spec-anchored legacy modernization for regulated industries — source-referenced rule extraction, characterization testing, parity guarantees, and deterministic CI gates. Sits beside the build; never modifies application code.

Take a snapshot of what your code does, then see if your edit moved anything. For code with no tests. Zero dependencies, MIT.

Lock observable software behavior before refactors, migrations, upgrades, or AI-assisted changes.

Homebrew tap for golden-census: brew install bunkanoyume/golden-census/golden-census

Text-first characterization testing for Swift — golden snapshots (JSON / view hierarchy / accessibility tree) with fixture support, test-target only, zero shipping footprint.

Differential-testing harness that verified a Django 4.2 -> 5.2 upgrade of a 9k-star SaaS: 770 recorded HTTP request/response pairs, replayed on both versions and compared byte-for-byte.

Esercizio L30 – AI Coding Upskilling: test di caratterizzazione API e unitari per proteggere il comportamento osservabile di una dashboard di ticket.

Safe, verified legacy code modernization with IBM Bob 2.0: tests first, then upgrade, then prove nothing broke.

GenAI-assisted, verification-driven modernization of mailgun/flanker: removes the Python 2 compat layer with byte-identical output proven via a characterization-test harness.

Watch AI migrate a legacy app to a modern stack, and prove it still behaves the same.

Refactor verificado con un test de equivalencia: dos funciones duplicadas en una docena de pantallas de un sistema de costos agrícolas en producción, unificadas y comprobadas contra el código anterior en 60.298 casos. TypeScript, cero dependencias.

A public, reproducible legacy modernization: Java 8 / Spring Boot 1.5 / AngularJS 1.8 to Java 25 / Spring Boot 4.1 / Angular 22 — safety net first, honest numbers, a measured AI test-generation experiment, and a German migration playbook. Both stands live.

Warsztat: przepisanie modułu C#/.NET na Java Spring Boot z zachowaniem zachowania (behaviour-preserving) i testami charakteryzującymi. Repo startowe — użyj 'Use this template'.

Golden and property-based tests for JSON codecs in Kotlin, with Jackson or kotlinx.serialization

Characterization tests and a living spec for legacy JavaScript: the real code, not the LLM, writes every expected value, and a mutation score shows whether the tests catch changes.