rvr ai

AI-powered commands for code review, interactive chat, code generation, and architecture design.

Usage

rvr ai <subcommand> [options]

Subcommands

chat

Start an interactive AI chat session scoped to your project context.

rvr ai chat
rvr ai chat --provider claude
rvr ai chat --provider ollama --model codellama

generate

Generate code from a natural language description.

rvr ai generate "Create a payment service with Stripe integration"
rvr ai generate "Add email notification when order status changes" --provider openai

review

Run AI-powered code analysis on your project.

# Run all analyzers
rvr ai review --all

# Run specific analyzers
rvr ai review --architecture    # Clean Architecture conformance
rvr ai review --ddd             # DDD anti-pattern detection
rvr ai review --performance     # N+1 queries, missing async, EF anti-patterns
rvr ai review --security        # OWASP vulnerability scanning

design

Launch an interactive architecture design assistant.

rvr ai design
rvr ai design --provider openai

Providers

Provider	Flag	Description
OpenAI	--provider openai	GPT-4 and GPT-3.5 models
Claude	--provider claude	Anthropic Claude models
Ollama	--provider ollama	Local/offline models

Configure the default provider in rivora.json:

{
  "ai": {
    "defaultProvider": "claude",
    "openai": { "apiKey": "${OPENAI_API_KEY}" },
    "claude": { "apiKey": "${CLAUDE_API_KEY}" },
    "ollama": { "endpoint": "http://localhost:11434" }
  }
}

Review Options

Flag	Description
--all	Run all analyzers
--architecture	Check Clean Architecture conformance
--ddd	Detect DDD anti-patterns
--performance	Find performance issues
--security	Scan for OWASP vulnerabilities
--output <format>	Output format: console, json, sarif
--output-file <path>	Write results to file
--severity <level>	Minimum severity: info, warning, error

SARIF Output for CI/CD

Generate SARIF reports for integration with GitHub Code Scanning or Azure DevOps:

rvr ai review --all --output sarif --output-file review.sarif

Use in a GitHub Actions workflow:

- name: AI Review
  run: rvr ai review --all --output sarif --output-file results.sarif

- name: Upload SARIF
  uses: github/codeql-action/upload-sarif@v3
  with:
    sarif_file: results.sarif

Examples

Offline review with Ollama:

rvr ai review --all --provider ollama

Generate code and review it:

rvr ai generate "Create a notification aggregate with email and SMS channels"
rvr ai review --architecture --ddd