@oddessentials/odd-ai-reviewers

AI-powered code review CLI that runs locally or in CI/CD pipelines. Get instant feedback on your code changes using LLMs like GPT-4 and Claude.

Quick Start

# Run local review on current changes
npx @oddessentials/odd-ai-reviewers .

# Or install globally
npm install -g @oddessentials/odd-ai-reviewers
ai-review .

Features

• Local Review Mode: Get instant AI feedback before pushing
• CI/CD Integration: GitHub Actions and Azure DevOps support
• Zero Configuration: Works out of the box with sensible defaults
• Multiple AI Providers: OpenAI, Anthropic, Azure OpenAI, Ollama
• Cost Control: Built-in budget limits and cost estimation
• Customizable: Configure passes, agents, and output formats

Prerequisites

• Node.js >= 22.0.0
• Git
• API key for your preferred AI provider

Environment Variables

Set one of these API keys:

# OpenAI
export OPENAI_API_KEY=sk-...

# Anthropic
export ANTHROPIC_API_KEY=sk-ant-...

# Azure OpenAI
export AZURE_OPENAI_API_KEY=...
export AZURE_OPENAI_ENDPOINT=https://...
export AZURE_OPENAI_DEPLOYMENT=...

# Ollama (local)
export OLLAMA_BASE_URL=http://localhost:11434

Usage

Local Review (Recommended for Development)

# Review uncommitted changes
ai-review .

# Review only staged changes (for pre-commit hooks)
ai-review . --staged

# Review changes between current branch and main
ai-review . --base main

# Dry run - see what would be reviewed
ai-review . --dry-run

# Cost estimate only
ai-review . --cost-only

# JSON output for tooling integration
ai-review . --format json

# SARIF output for IDE integration
ai-review . --format sarif

CI Review Mode

# Review a specific commit range
ai-review review --repo . --base $BASE_SHA --head $HEAD_SHA

# With PR number for GitHub integration
ai-review review --repo . --base $BASE_SHA --head $HEAD_SHA --pr 123

Configuration

# Generate a new configuration file
ai-review config init

# Validate existing configuration
ai-review validate --repo .

Configuration File

Create .ai-review.yml in your repository root:

# AI Provider configuration
provider: openai

# Model selection
models:
  default: gpt-4o-mini

# Review passes
passes:
  - name: ai-review
    enabled: true
    agents:
      - opencode

# Resource limits
limits:
  max_files: 50
  max_diff_lines: 5000
  max_usd_per_pr: 0.50

# Gating (fail CI on findings)
gating:
  enabled: true
  fail_on_severity: error

CLI Reference

ai-review local [path]

Run AI review on local changes.

| Option | Description | | ----------------- | --------------------------------------------- | | --base <ref> | Base reference for comparison (auto-detected) | | --head <ref> | Head reference (default: HEAD) | | --range <range> | Git range (e.g., HEAD~3..) | | --staged | Review only staged changes | | --uncommitted | Include uncommitted changes (default: true) | | --pass <name> | Run specific pass only | | --agent <id> | Run specific agent only | | --format <fmt> | Output format: pretty, json, sarif | | --no-color | Disable colored output | | --quiet | Minimal output (errors only) | | --verbose | Show debug information | | --dry-run | Preview without running agents | | --cost-only | Estimate cost without running | | -c, --config | Path to config file |

Exit Codes

| Code | Meaning | | ---- | ---------------------------------------------- | | 0 | Success (no blocking findings) | | 1 | Failure (blocking findings or execution error) | | 2 | Invalid arguments or configuration |

Pre-commit Hook Integration

Add to .pre-commit-config.yaml:

repos:
  - repo: local
    hooks:
      - id: ai-review
        name: AI Code Review
        entry: npx @oddessentials/odd-ai-reviewers . --staged --quiet
        language: system
        pass_filenames: false

Or use with Husky:

# .husky/pre-commit
npx @oddessentials/odd-ai-reviewers . --staged --quiet || exit 1

Changelog

This package uses semantic-release for automated versioning and changelog generation. The changelog is auto-generated from conventional commit messages.

See CHANGELOG.md for the full release history.

License

MIT