Add a CI gate that fails pull requests carrying strong AI generated code tells.
Run a pre commit hook that checks only staged changes for slop patterns.
Audit an entire source tree once for lingering AI generated artifacts.
Tune which rules block the build versus which only produce a warning.
| cj-vana/unslop-ci | 0xradioac7iv/tempfs | 52191314/web-agent-proxy-sdk | |
|---|---|---|---|
| Stars | 0 | 0 | 0 |
| Language | TypeScript | TypeScript | TypeScript |
| Setup difficulty | easy | moderate | moderate |
| Complexity | 3/5 | 3/5 | 4/5 |
| Audience | developer | developer | developer |
Figures from each repo's GitHub metadata at analysis time.
Runs via npx with no install required for a quick check, or npm install for CI and pre-commit integration.
unslop-ci is a continuous integration tool written in TypeScript that checks pull requests for the telltale signs of AI generated content, sometimes called AI slop, and can fail a build when it finds strong ones. What sets it apart from a general code linter is that it is diff aware: it only looks at the lines a change actually adds, not the whole file or the whole repository, so you can turn it on for an old codebase without first cleaning up everything that came before. Three separate scanners run over added lines depending on the file type. The code scanner covers languages like TypeScript, Python, Go, and SQL, and looks for things like leftover chat assistant phrases, placeholder stubs such as a comment saying rest of your code, swallowed errors like an empty catch block, emoji in source code, and generic function names. The text scanner covers markdown and similar files, catching patterns like the em dash, the not just X it's Y sentence structure, assistant style sign offs, and certain overused word choices. The ui scanner covers component and style files, flagging things like untouched default component library styling, a common purple and indigo AI color scheme, and gradient heading text. The rules are ported from an existing open source project and are grounded in a large scale analysis of what people actually flag as AI giveaways online. You can run it from the command line with npx, wire it into GitHub Actions as a workflow step, or hook it into git as a pre commit check so staged changes are checked before you commit. Findings are grouped by severity, high, medium, or low, and by whether they represent an actual bug versus just a stylistic tell. The default gate blocks high severity findings and anything classified as a bug, while everything else shows up only as a warning. You can adjust this threshold, exclude paths, disable individual rules, or mark a specific line as intentional with an inline comment so the gate does not fight you on deliberate choices. Configuration lives in an unslop.config.json file at the repository root. The project is released under the MIT license.
A diff-aware CI tool that scans only the lines a pull request adds for signs of AI generated code, text, and UI, and can block the build.
Mainly TypeScript. The stack also includes TypeScript, Node.js, GitHub Actions.
Setup difficulty is rated easy, with roughly 5min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
double-check against the repo, no cap.