vibescore

🎵 vibescore

Grade your vibe-coded project. One command. Instant letter grade.

"Vibe coding" is the new reality — you prompt, AI writes, you ship.
But is your vibe-coded project actually good?
Find out in 10 seconds.

$ vibescore .

🎵 Vibe Check  v0.1.0
══════════════════════════════════════════════

  Project:   tokonomics
  Files:     40 (32 Python, 8 other)
  Lines:     4,658
  Scanned in 0.12s

┌──────────────────┬────────┬───────┐
│ Category         │ Score  │ Grade │
├──────────────────┼────────┼───────┤
│ Code Quality     │   52.0 │ F     │
│ Security         │  100.0 │ A+    │
│ Dependencies     │   98.0 │ A+    │
│ Testing          │  100.0 │ A+    │
├──────────────────┼────────┼───────┤
│ Overall          │   87.6 │ B+    │
└──────────────────┴────────┴───────┘

🟡 Warnings (11)
  VC201  Function 'export_svg_chart' too long (102 lines)
  VC202  Function '_build_cli' high complexity (30)
  VC203  Function 'export_svg_chart' has 6 parameters (>5)
  ...

💡 Tips
  • Reduce function complexity and add type annotations

Install

pip install vibescore

That's it. Zero dependencies. Works with Python 3.9+.

Usage

# Grade the current directory
vibescore .

# Grade a specific project
vibescore /path/to/project

# JSON output (for CI pipelines)
vibescore . --format json

# Fail CI if score is below threshold
vibescore . --min-score 70

As a Python library

from vibescore import scan

report = scan(".")
print(f"Grade: {report.overall_grade} ({report.overall_score:.0f}/100)")

for category in report.categories:
    print(f"  {category.name}: {category.grade}")

What It Checks

Grading Scale

CI Integration

GitHub Actions

- name: Vibe Check
  run: |
    pip install vibescore
    vibescore . --min-score 70

Pre-commit (manual)

# In your Makefile or CI script
vibescore . --min-score 70 --format json > vibe-report.json

Pre-commit

repos:
  - repo: https://github.com/stef41/vibescore
    rev: v0.1.0
    hooks:
      - id: vibescore
        args: ["--min-score", "70"]

How Scoring Works

Each category is scored 0–100 independently. The overall score is a weighted average:

Security is weighted highest because a security bug in vibe-coded projects can be catastrophic.

Why vibescore?

Vibe coding means AI writes most of your code. That's fast, but it introduces risks:

• AI hallucinates long functions that are hard to debug
• AI skips security basics like input validation and secret management
• AI often omits tests or writes superficial ones
• AI uses loose dependency pins that break on updates

vibescore catches these patterns in seconds, so you can ship fast and ship safe.

FAQ

Q: Does this only work with Python?
A: Currently Python-focused for code quality and testing analysis. Security and dependency checks work with any project type. More languages coming soon.

Q: Does it phone home or require an API key?
A: No. Zero network requests. Zero dependencies. Runs entirely offline.

Q: How is this different from pylint/ruff/flake8?
A: Those are line-level linters. vibescore gives you a project-level grade across security, quality, testing, and dependencies — a holistic view of your vibe-coded project's health. Use both.

See Also

Tools in the same ecosystem:

• tokonomics — LLM token cost management
• injectionguard — Prompt injection detection
• vibesafe — AI code safety scanner
• castwright — Synthetic training data generator
• infermark — LLM inference benchmarking

License

Apache-2.0