npx, bunx, pnpm dlx, uvx: Run aislop scan Without Installing

Run npx aislop scan from the project root to score a repository 0 to 100 and list AI slop findings by file. It needs no install, account or model key. aislop is also published to PyPI and Homebrew, and every channel runs the same CLI, so use the runner your project already has.

Pick your runner

The first four run the package once and leave nothing installed globally. pipx and Homebrew install an aislop command you can rerun with aislop scan.

npx aislop scan

npx aislop@latest scan

@latest makes npx check for the current release instead of reusing a cached copy. Drop it if you want the cached version.

bunx aislop scan

bunx aislop scan

bunx fetches aislop from the npm registry. The CLI declares a Node shebang, which Bun respects, so keep Node.js 20 or later installed alongside Bun. To install it globally instead: bun add -g aislop.

pnpm dlx and yarn dlx

pnpm dlx aislop scan
yarn dlx aislop scan

Both are one-off runs, like npx. Yarn 2+ (Berry) has no global add, so yarn dlx is the way to run it there.

uvx and pipx for Python projects

uvx aislop scan
pipx install aislop

The aislop package on PyPI is a launcher that pins the matching npm release, so Python teams get the same CLI and score. It still needs Node.js on PATH because the engines run on Node. Run aislop doctor once to confirm everything is found.

Homebrew

brew install scanaislop/tap/aislop
aislop scan

The formula lives in the scanaislop/homebrew-tap repository.

Useful flags

• aislop scan --changes scans only what changed against HEAD. Add --base origin/main to compare against a branch.
• aislop scan --staged scans staged files, which suits a pre-commit hook.
• aislop scan --json prints machine-readable output for CI and scripts.

Flags work the same with every runner. The full list is in docs/commands.

How the score works

aislop weights findings from the format, lint, code quality, AI slop and security engines into one number, so style issues cost less than swallowed exceptions or missing imports. 75 to 100 is Healthy, 50 to 74 Needs Work, and below 50 is Critical. Details: scoring. For what findings look like, see 20 AI slop code examples.

Fix and CI

aislop fix applies mechanical cleanup, and aislop fix --claude, --cursor or --codex sends the remaining findings back to the agent that wrote them.

For a merge gate, aislop init writes config and a GitHub Actions workflow, and aislop ci fails the build below your threshold. Pin the version in CI so the score only moves when the code does. See CI docs and add a quality gate in 2 minutes.

Source: github.com/scanaislop/aislop. Product: what aislop is. No terminal handy? Scan a public repo in the browser.