yuvrajangadsingh/vibecheck
ESLint for AI slop. Detect AI-generated code smells in your JS/TS and Python codebase.
What's novel
ESLint for AI slop. Detect AI-generated code smells in your JS/TS and Python codebase.
Code Analysis
13 files read · 4 roundsA zero-config static analysis CLI that detects AI-generated code smells (security vulnerabilities, error-handling anti-patterns, AI tells, framework issues) in JavaScript/TypeScript and Python using regex rules, a hand-written lexer, and multiline structural analysis, with diff-aware scanning, calib
Strengths
The hand-written lexer with proper state machine handling of strings, templates, regexes, and comments is genuinely non-trivial and well-engineered; the scoring system is calibrated against a real corpus with thoughtful design decisions (per-rule caps, exponential decay, severity weighting) documented with the reasoning behind each choice. The multiline rules (empty catch detection, console pollution with dev-guard tracking, Python handler body scanning) show real algorithmic depth beyond simple
Weaknesses
The core detection mechanism is still fundamentally regex-based, which limits precision for complex code patterns and will inevitably produce false positives/negatives on edge cases; the hallucinated imports list is a static curated list that will go stale as AI models evolve and new hallucinated package names appear.
Score Breakdown
Signal breakdown
Innovation
Craft
Traction
Scope
Evidence
Commits
13
Contributors
1
Files
35
Active weeks
3
Repository
Language
TypeScript
Stars
5
Forks
0
License
MIT