codivus 0.1.0

AI-powered Python library for automated code reviews

Codivus

Codivus is an AI-assisted automated code review engine designed for Python codebases. It operates by combining deterministic Abstract Syntax Tree (AST) static analysis with Large Language Models (LLMs) to perform thorough, context-aware code audits, security scanning, and style checking.

---

Key Features

• Hybrid Review Engine: Merges traditional parsing-based static rules with advanced LLM reasoning to ensure zero false positives on basic rules while gaining deep semantic feedback.
• Provider Abstraction: Out-of-the-box integrations for OpenAI, Anthropic Claude, Google Gemini, Ollama, OpenRouter, and Azure OpenAI.
• Execution Caching: Smart local SHA-256 caching saves execution runtimes and limits API cost by ignoring unchanged files.
• Extensible Plugins: Programmatic hooks support custom prompt modifiers, pre-/post-review lifecycle callbacks, and custom static analysers.
• Git Diffs Integration: Compare branches, commit ranges, or analyze unstaged/staged git files.
• Multiple Report Formats: Export findings as Markdown, HTML, JSON, or SARIF schemas.
• CLI Commands: Interactive terminal command suite to run reviews, query settings, and explain code issues.

---

Installation

Install directly from PyPI:

pip install codivus

LLM Client Extras

To install specific provider client SDK packages:

# OpenAI support (default)
pip install codivus

# Anthropic Claude support
pip install codivus[anthropic]

# Google Gemini support
pip install codivus[google]

# Install all SDK libraries
pip install codivus[all-providers]

---

Quickstart

Programmatic Usage

Create a script run_review.py to review a python file:

import os
from codereview.config import Config
from codereview.reviewer import Reviewer

# Load configuration (uses defaults or environment settings)
config = Config()

# Instantiate the reviewer
reviewer = Reviewer(config)

# Execute the review on a file
result = reviewer.review_file("src/main.py")

# Output metrics
print(f"Overall Quality Score: {result.score.overall_score}/100")
print(f"Total Issues Found: {result.summary.total_issues}")

for issue in result.issues:
    print(f"[{issue.severity.upper()}] Line {issue.line_number}: {issue.title}")
    if issue.suggestion:
        print(f"  Proposed Code:\n{issue.suggestion.proposed_code}")

CLI Command Suite

# Review a single file
codivus review main.py

# Review a directory recursively
codivus repo review src/

# Run a dedicated security audit
codivus security scan main.py

# Explain a code snippet using AI
codivus explain "def hello(): pass"

---

Configuration

Set environment variables in your terminal or a local .env file:

# Select default provider client (openai, anthropic, google, ollama, openrouter, azure)
export CODIVUS_PROVIDER="openai"

# Configure the target model
export CODIVUS_MODEL="gpt-4o"

# Disable review caching (default is enabled)
export CODIVUS_NO_CACHE=0

# Configure logger level (DEBUG, INFO, WARNING, ERROR)
export CODIVUS_LOG_LEVEL="WARNING"

---

Documentation

Full tutorials, design architectures, and API references are available in the docs/ directory or can be hosted via MkDocs:

# Install mkdocs and theme
pip install mkdocs-material

# Start the local docs server
mkdocs serve