quality
Built a fast, predictable code-quality CLI and GitHub Action that detects and coordinates native analyzers across Rust, Swift, Android, Kotlin, JavaScript, and Astro repositories.
En esta página 1 sección
Project notes
Active project
One quality workflow for polyglot repositories
- Estimated reading time
- About 3 minutes
- In this project story
- 6 sections
I built quality for repositories where “run the checks” stops being a simple instruction.
A modern product may contain Rust services, a Swift application, Android modules, TypeScript packages, and an Astro site. Each ecosystem already has good analyzers. The missing layer is a predictable way to detect them, run them together, normalize the results, and explain what a repository expects.
quality is that layer. It coordinates native tools rather than replacing them.
Outcomes
Coordinates 13 native analyzers across Rust, Swift, Android, Kotlin, JavaScript, TypeScript, and Astro
Supports changed-file runs, repository baselines, external adapters, concurrent execution, and cross-repository audits
Ships native binaries and a GitHub Action with annotations, job summaries, checksum verification, and SARIF output
Goals
Preserve ecosystem expertise by invoking the analyzers teams already trust.
Give every repository one memorable workflow for checking, formatting, fixing, diagnosing, and adopting policy.
Make gradual adoption practical with explicit configuration and baselines for existing findings.
Use the same policy locally and in CI with structured output for people, agents, and GitHub.
What I built
A native Rust CLI
with
init,doctor,check,format,fix,baseline,repositories,instructions,completions, and CI-generation commands.Thirteen built-in adapters
covering Cargo fmt, Clippy, SwiftLint, SwiftFormat, Android Lint, detekt, ktlint, ESLint, Astro Check, Prettier, CSpell, Knip, and Actionlint.
Repository task discovery
that preserves canonical package scripts and monorepo-specific type-check or validation semantics.
Changed-file execution
that sends relevant paths to file-capable tools while escalating configuration changes to full checks.
Baseline support
that records existing diagnostics and fails only for new regressions without hiding missing tools or analyzer crashes.
A GitHub Action
with verified downloads, pull-request annotations, job summaries, SARIF reports, and configurable failure thresholds.
Technical highlights
- Concurrent execution
independent checks run together by default, with
--fail-fastavailable when the first failure matters more than a complete report.- Normalized diagnostics
pretty, JSON, SARIF, and GitHub output formats turn different analyzer conventions into one result model.
- Custom adapters
teams can add organization-specific tools with explicit commands, file modes, extensions, configuration files, and parsers.
- Cross-repository operations
audit a folder of repositories, preview missing policy, and apply configuration from one command surface.
- Portable releases
checksum-verified binaries target macOS, Linux, and Windows, alongside the versioned GitHub Action.
Results
- One entry point for polyglot quality checkswhile each ecosystem keeps its strongest analyzer.
- Faster feedbackthrough concurrent and changed-file-aware execution.
- Incremental enforcementthat lets established repositories block new regressions before paying down every old finding.
- Consistent evidence in pull requeststhrough annotations, summaries, SARIF, and explicit exit behavior.
Why it matters
Quality policy becomes fragile when it lives in tribal knowledge, a collection of unrelated scripts, or a CI file nobody runs locally.
It becomes heavy when a wrapper tries to reimplement every analyzer.
quality takes a narrower path: detect repository intent, resolve native tools reproducibly, run them efficiently, and make their results understandable everywhere the work happens.
De este proyecto
Dentro del trabajo.
Engineering notes
One quality command across polyglot repositories
How I built a native CLI that coordinates existing analyzers across Rust, Swift, Android, Kotlin, JavaScript, and Astro without flattening their ecosystem-specific strengths.
Leer artículoTool ecosystem
How I Built the santi020k Developer Tool Ecosystem
How Lumen, ESLint Config Basic, Quality, Astro Doctor, Commitprompt, Dep Beacon, OG, Theme, and Auth form one practical system for my projects.
Leer artículoComparte este artículo
Compártelo.
La copia automática no está disponible.
