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.

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.

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.

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.

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

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

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

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

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

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.

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.

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.

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

Prove your port behaves like the original. Agent skill for Claude Code & Codex + zero-dependency runner: run legacy and new code on the same inputs, shrink mismatches to minimal cases, get an evidence report. C# → Python, Java → Kotlin, any language pair.