New:Introducing Socket Scanning for VS Code Marketplace Extensions.Learn more →
Get Started

@agents-npm-packages/turbolint

Package Overview
Dependencies
Maintainers
1
Versions
1
Alerts
File Explorer

Advanced tools

Socket logo

Install Socket

Detect and block malicious and high-risk dependencies

Install

@agents-npm-packages/turbolint

Ultra-fast JavaScript/TypeScript linter CLI

latest
Source
npmnpm
Version
0.1.0
Version published
Maintainers
1
Created
Source

TurboLint

Ultra-fast JavaScript/TypeScript linter CLI implemented in Rust on top of Oxc.

Install

npm install @agents-npm-packages/turbolint

Then run the published binary:

turbolint ./src

Build

cargo build --release

Run

./target/release/turbolint ./src

The npm package ships the same command name and falls back to a local Cargo build if a native binary is not already present.

TurboLint recursively scans .js, .jsx, .ts, and .tsx files while ignoring node_modules and .git. It exits with 1 when violations are found and 0 when the tree is clean.

Options

  • --config PATH: load rule toggles and extra ignore patterns from a JSON file
  • --ignore-file PATH: load extra ignore globs from a text file
  • --json: emit structured JSON output with scan stats and violations
  • --fix: apply safe autofixes for no-var by rewriting var to let

Website

The repo also ships a static marketing/docs site in site/ with dark shadcn-inspired styling, command examples, and a product overview. The live hostname is turbolint.ideatr.dev; GitLab hosts the public source and docs, while Vercel serves the live site.

Rules

  • no-eval: flags eval(...)
  • no-var: flags var declarations
  • no-console-log: flags console.log(...)

Only no-var is auto-fixable. The other rules are report-only by design.

Architecture

  • Streaming directory traversal: paths are pushed through a bounded channel, avoiding full-tree materialization.
  • Fixed worker pool: worker count is bounded by logical CPU availability.
  • Memory-mapped file reads: source files are mapped read-only and parsed without copying the entire file into a new buffer.
  • Parallel directory crawling: traversal uses the ignore crate's parallel walker instead of a single recursive loop.
  • Per-file memory lifetime: source text, line index, Oxc allocator, AST, and rule state are dropped immediately after each file is processed.
  • Bounded queues: the path and result queues provide backpressure under very large repositories to keep memory usage predictable.
  • Single-pass file counting: the directory walker records scanned files while streaming paths to workers.

Deployment

  • Public source repo: gitlab.com/arjunkshah/turbolint
  • Live site: turbolint.ideatr.dev
  • Docs: site/docs/

FAQs

Package last updated on 16 Jul 2026

Related posts