openrabbit/README.md

OpenRabbit

Enterprise-grade AI code review system for Gitea and GitHub with automated PR review, issue triage, interactive chat, and codebase analysis.

---

Features

Feature	Description
PR Review	Inline comments, security scanning, severity-based CI failure
PR Summaries	Auto-generate comprehensive PR summaries with change analysis and impact assessment
Issue Triage	On-demand classification, labeling, priority assignment via @codebot triage
Chat	Interactive AI chat with codebase search and web search tools
@codebot Commands	@codebot summarize, changelog, explain-diff, explain, suggest, triage, review-again in comments
Codebase Analysis	Health scores, tech debt tracking, weekly reports
Security Scanner	17 OWASP-aligned rules + SAST integration (Bandit, Semgrep)
Dependency Scanning	Vulnerability detection for Python, JavaScript dependencies
Test Coverage	AI-powered test suggestions for untested code
Architecture Compliance	Layer separation enforcement, circular dependency detection
Notifications	Slack/Discord alerts for security findings and reviews
Compliance	Audit trail, CODEOWNERS enforcement, regulatory support
Multi-Provider LLM	OpenAI, Anthropic Claude, Azure OpenAI, Google Gemini, Ollama
Enterprise Ready	Audit logging, metrics, Prometheus export
Gitea Native	Built for Gitea workflows and API (also works with GitHub)

---

📦 Installation

Quick Setup (5 minutes):

# Clone OpenRabbit
git clone https://github.com/YourOrg/openrabbit.git
cd openrabbit

# Run interactive setup wizard
./setup.sh

The wizard will generate workflow files, create configuration, and guide you through the remaining steps.

📖 See INSTALL.md for:

• Detailed installation instructions
• Manual setup guide
• Platform-specific differences (Gitea vs GitHub)
• Troubleshooting common issues

---

Quick Start

1. Set Repository/Organization Secrets

AI_PROVIDER         - LLM provider: openai | openrouter | ollama | anthropic | azure | gemini
AI_MODEL            - Model to use for the active provider (e.g. gpt-4.1-mini, claude-3-5-sonnet-20241022)
OPENAI_API_KEY      - OpenAI API key (or use OpenRouter/Ollama)
SEARXNG_URL         - (Optional) SearXNG instance URL for web search

For Gitea:

AI_REVIEW_TOKEN     - Bot token with repo + issue permissions

For GitHub: The built-in GITHUB_TOKEN is used automatically.

2. Add Workflows to Repository

Workflows are located in .gitea/workflows/.

Gitea Example

Gitea PR Review Workflow

# .gitea/workflows/enterprise-ai-review.yml
name: AI PR Review
on: [pull_request]

jobs:
  ai-review:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
        with:
          fetch-depth: 0

      - uses: actions/checkout@v4
        with:
          repository: YourOrg/OpenRabbit
          path: .ai-review
          token: ${{ secrets.AI_REVIEW_TOKEN }}

      - uses: actions/setup-python@v5
        with:
          python-version: "3.11"

      - run: pip install requests pyyaml

      - name: Run AI Review
        env:
          AI_REVIEW_TOKEN: ${{ secrets.AI_REVIEW_TOKEN }}
          AI_REVIEW_REPO: ${{ gitea.repository }}
          AI_REVIEW_API_URL: https://your-gitea.example.com/api/v1
          AI_PROVIDER: ${{ secrets.AI_PROVIDER }}
          AI_MODEL: ${{ secrets.AI_MODEL }}
          OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}
        run: |
          cd .ai-review/tools/ai-review
          python main.py pr ${{ gitea.repository }} ${{ gitea.event.pull_request.number }}

See .gitea/workflows/ for all workflow examples.

3. Create Labels (Automatic Setup)

Option A: Automatic Setup (Recommended)

Create an issue and comment:

@codebot setup-labels

The bot will automatically:

• Detect your existing label schema (e.g., Kind/Bug, Priority - High)
• Map existing labels to OpenRabbit's auto-labeling system
• Create only the missing labels you need
• Follow your repository's naming convention

Option B: Manual Setup

Create these labels in your repository for auto-labeling:

• priority: critical, priority: high, priority: medium, priority: low
• type: bug, type: feature, type: question, type: documentation
• ai-approved, ai-changes-required, ai-reviewed

---

Project Structure

tools/ai-review/
├── agents/                 # Agent implementations
│   ├── base_agent.py       # Abstract base agent
│   ├── issue_agent.py      # Issue triage & @codebot commands
│   ├── pr_agent.py         # PR review with security scan
│   ├── codebase_agent.py   # Codebase health analysis
│   ├── chat_agent.py       # Interactive chat with tool calling
│   ├── dependency_agent.py # Dependency vulnerability scanning
│   ├── test_coverage_agent.py # Test coverage analysis
│   └── architecture_agent.py  # Architecture compliance checking
├── clients/                # API clients
│   ├── gitea_client.py     # Gitea REST API wrapper
│   ├── llm_client.py       # Multi-provider LLM client with tool support
│   └── providers/          # Additional LLM providers
│       ├── anthropic_provider.py  # Direct Anthropic Claude API
│       ├── azure_provider.py      # Azure OpenAI Service
│       └── gemini_provider.py     # Google Gemini API
├── security/               # Security scanning
│   ├── security_scanner.py # 17 OWASP-aligned rules
│   └── sast_scanner.py     # Bandit, Semgrep, Trivy integration
├── notifications/          # Alerting system
│   └── notifier.py         # Slack, Discord, webhook notifications
├── compliance/             # Compliance & audit
│   ├── audit_trail.py      # Audit logging with integrity verification
│   └── codeowners.py       # CODEOWNERS enforcement
├── utils/                  # Utility functions
│   ├── ignore_patterns.py  # .ai-reviewignore support
│   └── webhook_sanitizer.py # Input validation
├── enterprise/             # Enterprise features
│   ├── audit_logger.py     # JSONL audit logging
│   └── metrics.py          # Prometheus-compatible metrics
├── prompts/                # AI prompt templates
├── main.py                 # CLI entry point
└── config.yml              # Configuration

.github/workflows/          # GitHub Actions workflows
├── ai-review.yml           # PR review workflow
├── ai-issue-triage.yml     # Issue triage workflow
├── ai-codebase-review.yml  # Codebase analysis
├── ai-comment-reply.yml    # @codebot command responses
└── ai-chat.yml             # Interactive AI chat

.gitea/workflows/           # Gitea Actions workflows
├── enterprise-ai-review.yml
├── ai-issue-triage.yml
├── ai-codebase-review.yml
├── ai-comment-reply.yml
└── ai-chat.yml

---

CLI Commands

# Review a pull request
python main.py pr owner/repo 123

# Triage an issue
python main.py issue owner/repo 456

# Respond to @codebot command
python main.py comment owner/repo 456 "@codebot explain"

# Analyze codebase
python main.py codebase owner/repo

# Chat with the AI bot
python main.py chat owner/repo "How does authentication work?"
python main.py chat owner/repo "Find all API endpoints" --issue 789

---

@codebot Commands

Issue Commands

In any issue comment:

Command	Description
@codebot help	Help: Show all available commands with examples
@codebot setup-labels	Setup: Automatically create/map repository labels for auto-labeling
@codebot triage	Full issue triage with auto-labeling and analysis
@codebot summarize	Summarize the issue in 2-3 sentences
@codebot explain	Explain what the issue is about
@codebot suggest	Suggest solutions or next steps
@codebot check-deps	Scan dependencies for security vulnerabilities
@codebot suggest-tests	Suggest test cases for changed code
@codebot refactor-suggest	Suggest refactoring opportunities
@codebot architecture	Check architecture compliance (alias: arch-check)
@codebot (any question)	Chat with AI using codebase/web search tools

Pull Request Commands

In any PR comment:

Command	Description
@codebot summarize	Generate a comprehensive PR summary with changes, files affected, and impact
@codebot changelog	Generate Keep a Changelog format entries ready for CHANGELOG.md
@codebot explain-diff	Explain code changes in plain language for non-technical stakeholders
@codebot review-again	Re-run AI code review on current PR state without new commits

PR Summary (@codebot summarize)

Features:

• 📋 Generates structured summary of PR changes
• ✨ Categorizes change type (Feature/Bugfix/Refactor/Documentation/Testing)
• 📝 Lists what was added, modified, and removed
• 📁 Shows all affected files with descriptions
• 🎯 Assesses impact scope (small/medium/large)
• 🤖 Automatically generates on PRs with empty descriptions

When to use:

• When a PR lacks a description
• To quickly understand what changed
• For standardized PR documentation
• Before reviewing complex PRs

Example output:

## 📋 Pull Request Summary
This PR implements automatic PR summary generation...

**Type:** ✨ Feature

## Changes
✅ Added:
- PR summary generation in PRAgent
- Auto-summary for empty PR descriptions

📝 Modified:
- Updated config.yml with new settings

## Files Affected
- ➕ `tools/ai-review/prompts/pr_summary.md` - New prompt template
- 📝 `tools/ai-review/agents/pr_agent.py` - Added summary methods

## Impact
🟡 **Scope:** Medium
Adds new feature without affecting existing functionality

Changelog Generator (@codebot changelog)

Features:

• 📋 Generates Keep a Changelog format entries
• 🏷️ Categorizes changes (Added/Changed/Fixed/Removed/Security)
• ⚠️ Detects breaking changes automatically
• 📊 Includes technical details (files, LOC, components)
• 📝 Ready to copy-paste into CHANGELOG.md

When to use:

• Preparing release notes
• Maintaining CHANGELOG.md
• Customer-facing announcements
• Version documentation

Example output:

## 📋 Changelog for PR #123

### ✨ Added
- User authentication system with JWT tokens
- Password reset functionality via email

### 🔄 Changed
- Updated database schema for user table
- Refactored login endpoint for better error handling

### 🐛 Fixed
- Session timeout bug causing premature logouts
- Security vulnerability in password validation

### 🔒 Security
- Fixed XSS vulnerability in user input validation

---

### ⚠️ BREAKING CHANGES
- **Removed legacy API endpoint /api/v1/old - migrate to /api/v2**

---

### 📊 Technical Details
- **Files changed:** 15
- **Lines:** +450 / -120
- **Main components:** auth/, api/users/, database/

Diff Explainer (@codebot explain-diff)

Features:

• 📖 Translates technical changes into plain language
• 🎯 Perfect for non-technical stakeholders (PMs, designers)
• 🔍 File-by-file breakdown with "what" and "why"
• 🏗️ Architecture impact analysis
• ⚠️ Breaking change detection
• 📊 Technical summary for reference

When to use:

• New team members reviewing complex PRs
• Non-technical reviewers need to understand changes
• Documenting architectural decisions
• Learning from others' code

Example output:

## 📖 Code Changes Explained (PR #123)

### 🎯 Overview
This PR adds user authentication using secure tokens that expire after 24 hours, enabling users to log in securely without storing passwords in the application.

### 🔍 What Changed

#### ➕ `auth/jwt.py` (new)
**What changed:** Creates secure tokens for logged-in users
**Why it matters:** Enables the app to remember who you are without constantly asking for your password

#### 📝 `api/users.py` (modified)
**What changed:** Added a login page where users can sign in
**Why it matters:** Users can now create accounts and access their personal data

---

### 🏗️ Architecture Impact
Introduces a security layer across the entire application, ensuring only authenticated users can access protected features.

**New dependencies:**
- PyJWT (for creating secure tokens)
- bcrypt (for password encryption)

**Affected components:**
- API (all endpoints now check authentication)
- Database (added user credentials storage)

---

### ⚠️ Breaking Changes
- **All API endpoints now require authentication - existing scripts need to be updated**

---

### 📊 Technical Summary
- **Files changed:** 5
- **Lines:** +200 / -10
- **Components:** auth/, api/

Review Again (@codebot review-again)

Features:

• ✅ Shows diff from previous review (resolved/new/changed issues)
• 🏷️ Updates labels based on new severity
• ⚡ No need for empty commits to trigger review
• 🔧 Respects latest .ai-review.yml configuration

When to use:

• After addressing review feedback in comments
• When AI flagged a false positive and you explained it
• After updating .ai-review.yml security rules
• To re-evaluate severity after code clarification

Example:

The hardcoded string at line 45 is a public API URL, not a secret.
@codebot review-again

New to OpenRabbit? Just type @codebot help in any issue to see all available commands!

Label Setup Command

The @codebot setup-labels command intelligently detects your existing label schema and sets up auto-labeling:

For repositories with existing labels (e.g., Kind/Bug, Priority - High):

• Detects your naming pattern (prefix/slash, prefix-dash, or colon-style)
• Maps your existing labels to OpenRabbit's schema
• Creates only missing labels following your pattern
• Zero duplicate labels created

For fresh repositories:

• Creates OpenRabbit's default label set
• Uses type:, priority:, and status labels

Example output:

@codebot setup-labels

✅ Found 18 existing labels with pattern: prefix_slash

Detected Categories:
- Kind (7 labels): Bug, Feature, Documentation, Security, Testing
- Priority (4 labels): Critical, High, Medium, Low

Proposed Mapping:
| OpenRabbit Expected | Your Existing Label | Status |
|---------------------|---------------------|--------|
| type: bug          | Kind/Bug            | ✅ Map |
| priority: high     | Priority - High     | ✅ Map |
| ai-reviewed        | (missing)           | ⚠️ Create |

✅ Created Kind/Question (#cc317c)
✅ Created Status - AI Reviewed (#1d76db)

Setup Complete! Auto-labeling will use your existing label schema.

---

Interactive Chat

The chat agent is an interactive AI assistant with tool-calling capabilities:

Tools Available:

• search_codebase - Search repository files and code
• read_file - Read specific files
• search_web - Search the web via SearXNG

Example:

@codebot How do I configure rate limiting in this project?

The bot will search the codebase, read relevant files, and provide a comprehensive answer.

---

Configuration

Edit tools/ai-review/config.yml:

# Set via AI_PROVIDER secret — or hardcode here as fallback
provider: openai   # openai | openrouter | ollama | anthropic | azure | gemini

# Set via AI_MODEL secret — or hardcode per provider here
model:
  openai: gpt-4.1-mini
  openrouter: anthropic/claude-3.5-sonnet
  ollama: codellama:13b

agents:
  issue:
    enabled: true
    auto_label: true
  pr:
    enabled: true
    inline_comments: true
    security_scan: true
  codebase:
    enabled: true
  chat:
    enabled: true
    searxng_url: ""  # Or set SEARXNG_URL env var

interaction:
  respond_to_mentions: true
  mention_prefix: "@codebot"  # Customize your bot name here!
  commands:
    - summarize
    - explain
    - suggest

---

Customizing the Bot Name

The default bot name is @codebot. To change it:

Step 1: Edit tools/ai-review/config.yml:

interaction:
  mention_prefix: "@yourname"  # e.g., "@assistant", "@reviewer", etc.

Step 2: Update all workflow files in .gitea/workflows/:

• ai-comment-reply.yml
• ai-chat.yml
• ai-issue-triage.yml

Look for and update:

if: contains(github.event.comment.body, '@codebot')

Change @codebot to your new bot name.

Step 3 (CRITICAL): Update bot username to prevent infinite loops:

In all three workflow files, find:

github.event.comment.user.login != 'Bartender'

Replace 'Bartender' with your bot's Gitea username. This prevents the bot from triggering itself when it posts comments containing @codebot, which would cause infinite loops and 10+ duplicate workflow runs.

---

Security Scanning

17 rules covering OWASP Top 10:

Category	Examples
Injection	SQL injection, command injection, XSS
Access Control	Hardcoded secrets, private keys
Crypto Failures	Weak hashing (MD5/SHA1), insecure random
Misconfiguration	Debug mode, CORS wildcard, SSL bypass

---

Documentation

Document	Description
Getting Started	Quick setup guide
Configuration	All options explained
Agents	Agent documentation
Security	Security rules reference
Workflows	GitHub & Gitea workflow examples
API Reference	Client and agent APIs
Enterprise	Audit logging, metrics
Troubleshooting	Common issues

---

LLM Providers

Provider	Model	Use Case
OpenAI	gpt-4.1-mini	Fast, reliable, default
Anthropic	claude-3.5-sonnet	Direct Claude API access
Azure OpenAI	gpt-4 (deployment)	Enterprise Azure deployments
Google Gemini	gemini-1.5-pro	GCP customers, Vertex AI
OpenRouter	claude-3.5-sonnet	Multi-provider access
Ollama	codellama:13b	Self-hosted, private

Provider Configuration

The provider and model can be set via Gitea secrets so you don't need to edit config.yml:

Secret	Description	Example
AI_PROVIDER	Which LLM provider to use	openrouter
AI_MODEL	Model for the active provider	google/gemini-2.0-flash

The config.yml values are used as fallback when secrets are not set.

# In config.yml (fallback defaults)
provider: openai  # openai | anthropic | azure | gemini | openrouter | ollama

# Azure OpenAI
azure:
  endpoint: ""  # Set via AZURE_OPENAI_ENDPOINT env var
  deployment: "gpt-4"
  api_version: "2024-02-15-preview"

# Google Gemini (Vertex AI)
gemini:
  project: ""  # Set via GOOGLE_CLOUD_PROJECT env var
  region: "us-central1"

Environment Variables

Variable	Provider	Description
AI_PROVIDER	All	Override the active provider (e.g. openrouter)
AI_MODEL	All	Override the model for the active provider
OPENAI_API_KEY	OpenAI	API key
ANTHROPIC_API_KEY	Anthropic	API key
AZURE_OPENAI_ENDPOINT	Azure	Service endpoint URL
AZURE_OPENAI_API_KEY	Azure	API key
AZURE_OPENAI_DEPLOYMENT	Azure	Deployment name
GOOGLE_API_KEY	Gemini	API key (public API)
GOOGLE_CLOUD_PROJECT	Vertex AI	GCP project ID
OPENROUTER_API_KEY	OpenRouter	API key
OLLAMA_HOST	Ollama	Server URL (default: localhost:11434)

---

Enterprise Features

• Audit Logging: JSONL logs with integrity checksums and daily rotation
• Compliance: HIPAA, SOC2, PCI-DSS, GDPR support with configurable rules
• CODEOWNERS Enforcement: Validate approvals against CODEOWNERS file
• Notifications: Slack/Discord webhooks for critical findings
• SAST Integration: Bandit, Semgrep, Trivy for advanced security scanning
• Metrics: Prometheus-compatible export
• Rate Limiting: Configurable request limits and timeouts
• Custom Security Rules: Define your own patterns via YAML
• Tool Calling: LLM function calling for interactive chat
• Ignore Patterns: .ai-reviewignore for excluding files from review

Notifications Configuration

# In config.yml
notifications:
  enabled: true
  threshold: "warning"  # info | warning | error | critical

  slack:
    enabled: true
    webhook_url: ""  # Set via SLACK_WEBHOOK_URL env var
    channel: "#code-review"

  discord:
    enabled: true
    webhook_url: ""  # Set via DISCORD_WEBHOOK_URL env var

Compliance Configuration

compliance:
  enabled: true
  audit:
    enabled: true
    log_file: "audit.log"
    retention_days: 90
  codeowners:
    enabled: true
    require_approval: true

---

License

MIT