Introducing aislop: the quality gate for AI-written code

#ai #aislop #codequality #cleancode

I got tired of reviewing pull requests that looked fine until they weren’t.

The code compiled.
The tests passed.
The feature worked on the surface.

But inside the diff, there would be things that should never have made it that far.

A silent catch {} swallowing every error.
A // TODO: implement this properly sitting in production code.
A dozen as any casts because the agent just forced the types through instead of wiring them properly.

That is the gap I kept seeing with AI-generated code.

Claude Code, Cursor, Copilot, Gemini, Codex — they are very good at producing code that satisfies the prompt. But satisfying the prompt and keeping a codebase healthy are not the same thing.

That gap compounds over time.

So I built aislop.

aislop is an open-source CLI and quality gate for teams using AI coding agents. It scans your codebase for the kind of patterns agents keep leaving behind. Not always bugs, but signals that the code was generated quickly without enough thought for maintainability.

Things like:

swallowed exceptions
unsafe as any casts
production TODOs
dead code
narrative comments
hardcoded URLs and config
fake fallbacks that hide real failures
duplicated logic from regenerated code

Each one looks small in isolation. But together, they create the kind of codebase that becomes harder to trust every sprint.

You can run it locally:

npx aislop scan

Or put it in CI so slop gets caught before it reaches main.

The reason I think this deserves its own tool is simple: normal linters were built for general code quality. aislop is built specifically around the patterns AI agents leave behind.

That is a different problem.

It also works with agent workflows. If you are using Claude Code, Cursor, Gemini, or similar tools, aislop can run as part of the agent loop instead of only after the code is done.

There is also an installable skill: