Skip to main content

QVC - AI Agent code quality external overseer. Review only, never modify.

Project description

QVC — QA of Vebe Coding

version python license tests rules languages

The external overseer for AI-generated code.
Finds what your AI Agent missed in its own review.
Never modifies a single line.

pip install qvc && qvc setup && qvc scan .

What makes QVC different

Tool Perspective AI Blind Spots Modifies Code
ESLint / Pylint Rules No Sometimes
SonarQube / CodeQL Enterprise SAST No No
AI Self-Review Internal (blind) No Yes
QVC External overseer Yes Never

When an AI Agent writes code and reviews it, it follows the same mental path both times. QVC comes from outside that loop — it sees what the Agent cannot.


Quick Demo

$ pip install qvc
$ qvc setup
$ qvc scan .

Scanning: ai-corp V7.1
Found 120 files, 19,705 lines
Static analysis... [====================] 100%

Report: qvc-report.md
Done! 1.0s, 2 bugs

[AHA] Your AI cannot see these:
  [90%] API signature mismatch (context loss blind spot)
  [85%] Silenced exception (boundary condition blind spot)

  AI missed rate: 67%
  Tasks: 2 -> .qvc/tasks/pending.md
  Say 'fix qvc tasks' to your AI.

Features

External Overseer Perspective

  • Blind spot classifier: Memory residue, context loss, self-review gaps
  • Self-review miss rate: Quantifies what your AI Agent missed
  • Cross-file consistency: Catches drift across multi-file AI generations

AI-Native Task Pool (V6)

  • Scan results auto-converted to AI-consumable tasks
  • Your AI Agent reads .qvc/tasks/pending.md and fixes issues
  • You just say: "fix qvc tasks"

Self-Evolving Gene Pool

  • 50 seed fingerprints from CWE Top 25
  • Learns from every scan — grows smarter locally
  • Community gene pool sync (optional, git-free fallback)

Works Everywhere

  • Zero config: No API keys, no GitHub, no setup
  • Offline-first: Core scanning fully offline
  • 6 languages: Python, JavaScript, TypeScript, React, Go, Universal
  • Any AI Agent: Codex, Cursor, GPT, Claude

Production Ready

  • 99.8% noise reduction: 2,000+ candidates -> 2-5 actionable issues
  • 42/45 tests passing, self-scan verified (0 real bugs)
  • 4 output formats: Markdown, JSON, SARIF, AI Fix Prompt
  • VS Code extension + GitHub Actions included

Installation

# One command
pip install qvc

# Or from source
git clone https://github.com/qvc-project/qvc.git
cd qvc && pip install -e .

One-click installers

  • Windows: Double-click install.bat
  • Mac/Linux: bash install.sh

Quick Start

qvc setup          # Initialize project (one time)
qvc scan .         # Full scan
qvc scan . --diff  # Scan changed files only
qvc watch .        # Auto-scan on file changes

Then say to your AI Agent: "fix qvc tasks"


Documentation

Document For
User Manual End users (Chinese)
Project Report Developers
Business Pitch Investors / Partners
Contributing Open source contributors

vs Alternatives

QVC ESLint SonarQube CodeQL
AI blind spot detection Yes No No No
Self-review gap analysis Yes No No No
AI Agent task integration Yes No No No
Zero config Yes Partial No No
Offline Yes Yes No No
Free / OSS MIT MIT Paid Open
Setup time 0 min 5 min 1 hour 30 min

QVC's self-scan

QVC V6.1 scanned itself:
  71 files / 7,875 lines / 0.5s
  Result: 0 real bugs found
  BOM: clean   Caches: clean   Log: clean

License

MIT — Free forever. Use it, fork it, ship it.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distributions

No source distribution files available for this release.See tutorial on generating distribution archives.

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

qvc_overseer-0.6.1-py3-none-any.whl (119.4 kB view details)

Uploaded Python 3

File details

Details for the file qvc_overseer-0.6.1-py3-none-any.whl.

File metadata

  • Download URL: qvc_overseer-0.6.1-py3-none-any.whl
  • Upload date:
  • Size: 119.4 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.13.2

File hashes

Hashes for qvc_overseer-0.6.1-py3-none-any.whl
Algorithm Hash digest
SHA256 4430bd46da93fa4787b22b4b98b09a595bc1322192834f03194e666853f0d6bf
MD5 aa705c65465600662ac5d1e63516b548
BLAKE2b-256 19b96b75c98ba8638c316db4961cb26b1c30d76b882158fd68a22ccc70e224e0

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page