code_buster 0.2.0

Deterministic offline repository architecture and multi-language static-analysis CLI.

[Code Buster Symbol]
Code Buster

Multi-language repository architecture and static-analysis CLI

About · Installation · Current Status · Documentation · Contributing

About

Code Buster reports dependency wiring, dead code, cycles, duplication, complexity, architecture-policy violations, feature flags, security and style heuristics, quality scores, and remediation plans.

This directory contains the canonical Dart implementation. Its executable is cb, its configuration file is code-buster.toml, and the workspace installer builds it from local source.

Is Code Buster a linter?

Not primarily. Code Buster includes lint-like correctness, security, and style findings, but its main scope is the repository rather than one source file. It builds views of dependencies, module boundaries, cycles, dead code, duplication, complexity, hotspots, and related files across supported languages.

A language-native linter usually has deeper knowledge of one language and reports local syntax, type, correctness, and style problems. Code Buster uses cross-file and cross-language evidence to answer broader questions such as:

• How is this repository connected?
• What might a change affect?
• Where are responsibilities duplicated or drifting?
• Which files or modules deserve investigation first?
• Is code bypassing an established architecture, component, or design token?

The tools are complementary. Continue using analyzers and linters such as dart analyze, ESLint, Clang-Tidy, or language-specific compiler diagnostics. Code Buster does not replace them.

	Language linter	Code Buster
Primary scope	A file, declaration, or expression	A repository and its relationships
Strongest evidence	Language syntax, types, and semantics	Dependency graphs, repeated structures, and cross-file patterns
Typical output	Direct diagnostics for a language	Repository views, advisory findings, and investigation paths
Language depth	Deep knowledge of one language	Mixed-language analysis with depth varying by language
Main use	Enforce established language rules	Understand unfamiliar code and guide human or AI-assisted changes

What is the aim?

The aim is to support the full change loop for developers and AI coding agents: understand the repository before editing, stay within its existing structures while editing, and inspect the result afterwards. An AI agent can run Code Buster against its own changes to find possible duplication, architecture violations, component drift, dead code, and other issues before handing work back for review.

Code Buster also helps keep repositories organized by making dependencies, boundaries, repeated implementations, and concentrated complexity visible over time. It should surface evidence and possible drift, not claim certainty that static analysis cannot provide. Before version 1.0.0, use it primarily for local investigation and non-blocking evaluation rather than as a production release gate.

Current Status

Overall status: 0.2.0 release candidate. The Dart implementation is the canonical runtime and passes strict analysis, the complete test suite, native compilation, documentation validation, self-analysis, and multi-repository precision checks.

Language status describes the current analysis depth.

Language	Status	Testing status	Current analysis evidence
Dart	High	Done	Analyzer AST, package-aware dependency graphs, function extraction, broad rule coverage, and self-hosting validation
C#	High	Needs more testing	Dedicated adapter suite with project-aware graph handling and broad correctness, reliability, security, and style rules
Java	High	Needs more testing	Dedicated adapter suite and focused regression suites for resources, exceptions, concurrency, SQL, cryptography, and package cycles
Nim	High	Needs more testing	Dedicated parser and rule packs with complete catalog wiring and focused regression coverage
Python	High	Needs more testing	Import and function extraction, graph analysis, dedicated adapter tests, and broad rule coverage
C/C++ and Objective-C	Moderate	Needs more testing	Dialect-aware source gating, includes, callable extraction, and targeted safety and modernization rules
Go	Moderate	Needs more testing	go.mod-aware imports, methods and functions, test classification, and focused reliability and security suites
JavaScript and TypeScript	Moderate	Needs more testing	Module graph and function extraction with frontend, Node.js, security, and TypeScript-specific checks
Lua and Luau	Moderate	Needs more testing	Module and callable extraction with targeted correctness, runtime, and style checks
SQL, PostgreSQL, and MySQL	Moderate	Needs more testing	Dialect-aware statement analysis with correctness, safety, and maintainability checks
Wren	Moderate	Needs more testing	Import and callable extraction with a dedicated rule pack and adapter regressions
CSS	Foundational	Needs more testing	Tested source discovery and targeted structural and style checks
HTML	Foundational	Needs more testing	Tested source discovery, embedded-script handling, and targeted correctness and style checks

High, Moderate, and Foundational describe implementation depth, not test completion. Done means the language has extensive unit, integration, self-hosting, and external-repository validation. Needs more testing means the existing automated coverage passes, but broader real-world validation is still required.

Installation

Code Buster requires Dart 3.11 or newer.

Homebrew (macOS)

Code Buster does not currently publish a Homebrew formula. Homebrew can install the required Dart SDK, after which Dart can install Code Buster directly from its Git repository:

brew tap dart-lang/dart
brew install dart
dart pub global activate --source git https://github.com/tool-bunker/code-buster.git
cb version

Ensure Dart's global executable directory is on PATH:

export PATH="$PATH:$HOME/.pub-cache/bin"

Local source build

dart pub get
dart compile exe bin/cb.dart -o build/cb
./build/cb version

Install the local checkout with ../install-code-buster.sh.