Catch the slop AI agents leave in your code. Score every PR.
Define your standard once. Every agent (Claude Code, Cursor, Codex) is held to it on every edit and every PR. 40+ rules, 8 languages, sub-second, no LLM. Free CLI, MIT-licensed.
The code your agent ships, and what aislop sees.
Eleven lines that compile, pass your tests, and read clean at a glance. aislop scores them in 0.4s, no LLM.
See your score in 10 seconds. No account needed.
Take any public GitHub URL and swap github.com for scanaislop.com. Or click any repo below to see it scanned right now.
Agentic skills framework and software development methodology that works
Toolkit to get started with Spec-Driven Development
Open-source multimodal AI agent stack for desktop automation
Skills for real engineers — straight from Matt Pocock's .claude directory
Implement a ChatGPT-like LLM in PyTorch from scratch, step by step
Open-source infrastructure for Computer-Use Agents — sandboxes and SDKs
Persistent memory layer for AI coding agents
Browser-based 3D Gaussian Splat editor
Lightning-fast, censorship-resistant proxy
AI browser automation framework for production workflows
Terminal-native AI coding agent built for real development loops
MCP server that gives agents structured browser automation
When AI-generated code moves faster than human review can keep up.
Define your bar once. The aislop CLI scans locally, scanaislop enforces on every PR. One score, one workflow, one standard, from first draft to merged pull request.
Wear the score on your README.
Free for any project. Drop it in your README and it shows your live aislop score. Green above 85, amber above 70, red below.
| 85+ | green | healthy |
| 70–84 | amber | debt |
| < 70 | red | failing the bar you set |
| -- | grey | not yet scanned publicly |
https://badges.scanaislop.com/score/<owner>/<repo>.svg [](https://scanaislop.com/<owner>/<repo>)
Public scans, published methodology.
aislop rules come from measured failure modes in real repositories, benchmark papers, and regression tests. The research hub tracks what we scanned, what changed, and which detector shipped.
Rule precision across seven language ecosystems
We scanned popular TypeScript, Python, Go, Rust, Ruby, PHP, and Java projects, then reduced noisy findings by 38% without disabling rules.
Read report →Benchmark signals converted into deterministic rules
SlopCodeBench measured verbosity and structural erosion. We translated the repeatable Python signals into four shipped rules.
Read report →First-run findings from real AI-assisted repositories
The earliest public scan run exposed the shapes that still drive the rule roadmap: narrative comments, dead code, unsafe casts, and broken fixers.
Read report →Guides and deep dives.
How-to guides, pattern breakdowns, and project updates.
aislop v0.9.4. SlopCodeBench called it verbosity. We turned it into rules.
Four new Python rules drawn from the verbosity signal in SlopCodeBench (SCBench, arXiv 2603.24755). Plus a CLI star prompt and GitHub Discussions for the community.
Read more →aislop v0.9.3. We measured the noise. Then we cut it by 38%.
Patch release focused on rule precision. Tightens detection across the ai-slop, security, lint, and source-file engines so language conventions are no longer flagged as slop. No new rules — existing ones now discriminate better.
Read more →A prompt is a suggestion. A CI gate is a guardrail. Ship the gate.
A prompt is a suggestion. A CI gate is a guardrail. Three lines of YAML and npx aislop@latest ci is the cheapest real guardrail you can ship today.
Read more →Get the monthly AI-code-quality research.
Each issue ties a public scan or benchmark to a shipped detector change. No fluff, one email a month.
Built in the open. Ready for you and your team.
MIT-licensed CLI. Sub-second, deterministic, no LLM at runtime. Same code in, same score out, every time. The hosted platform adds PR enforcement, team standards, and dashboards on top.
Built for teams standardizing Claude Code, Cursor, Codex, and CI