Skip to main content
aislop is the open-source CLI quality gate for AI generated code. Run it on a repo, get a score from 0 to 100, a file-and-line fix list, and a CI exit code you can enforce. scanaislop.com is the hosted platform around that CLI: badges, public scans, and team gates. These docs cover the CLI.
No config required. Same code in, same score out. There is no language model in the runtime path: aislop uses regex, AST analysis, and standard open-source tooling.

​
What it catches

AI coding agents (Claude Code, Cursor, Codex, OpenCode, Gemini CLI) often produce code that passes tests and lint, then leaves residue: narrative comments, swallowed exceptions, as any casts, hallucinated imports, dead code, TODO stubs, oversized functions. aislop encodes 50+ of those patterns and weights them so security and high-confidence defects hit harder than mechanical cleanup.

​
Six engines

Every scan runs six engines in parallel. Architecture is opt-in.

Formatting

Biome, ruff, gofmt, cargo fmt, rubocop, php-cs-fixer, dotnet format, clang-format.

Linting

oxlint, ruff, golangci-lint, clippy, expo-doctor, Roslynator / InspectCode, cppcheck / clang-tidy.

Code quality

Function and file size, nesting, unused files and dependencies via knip, unused declarations.

AI slop

Narrative and trivial comments, swallowed exceptions, as any, unused imports, TODO stubs, generic names.

Security

eval, innerHTML, injection patterns, hardcoded secrets, and dependency audits.

Architecture

Opt-in import bans, layering rules, and required patterns in .aislop/rules.yml.

​
10 language targets

Built-in checks still run when optional system tools are missing. If a repository is mostly an unsupported language (Swift, Kotlin, Java), aislop withholds the score and reports scoreable: false instead of scoring a handful of incidental files.

​
Why it is safe in CI

  • Deterministic. No LLM at runtime. The same tree always produces the same score.
  • Fast. Engines run in parallel. Most projects scan in well under a second.
  • Zero-config start. npx aislop@latest scan works on any supported repo. Add .aislop/config.yml when you want thresholds, rule overrides, or architecture rules.
  • Repair path. aislop fix handles mechanical cleanup. aislop agent runs a local Codex, Claude Code, or OpenCode session with aislop owning scoring and verification.
Current CLI: v0.16.0. MIT licensed.

​
Give this to your agent

Paste this into Claude Code, Cursor, Codex, or any other coding agent. It installs the aislop skill, hooks, and AGENTS.md standard.
Or run:
Full copy-paste prompt, AGENTS.md template, and URLs: For agents.

​
Next

Quickstart

Scan, read the score, repair, and add a CI gate.

Installation

npm, Homebrew, pipx, GitHub Packages, and optional tools.

How it works

Engines, language coverage, and scoring.

For agents

Copy a prompt or URL so the agent installs the skill, hooks, and AGENTS.md.