Command Reference

Complete reference for all Rawi commands, options, and capabilities.

Overview

Rawi (راوي) is a unified command-line tool to interact with multiple AI providers, manage chat sessions, and automate AI-powered workflows directly from your terminal.

Synopsis

rawi [global-options] <command> [command-options] [arguments]

Use rawi <command> --help for information on a specific command.

Global Options

Option	Type	Description
--debug	boolean	Turn on debug logging.
--profile <profile>	string	Use a specific profile from your configuration.
--output <format>	string	The formatting style for command output. Options: json, text, table, yaml, yaml-stream
--region <region>	string	The region to use (for cloud providers). Overrides config/env settings.
--color <mode>	string	Turn on/off color output. Options: on, off, auto
--version		Display the version of this tool.
--no-paginate	boolean	Disable automatic pagination for commands that support it.
--no-cli-pager	boolean	Disable CLI pager for output.
--cli-auto-prompt	boolean	Automatically prompt for CLI input parameters.
--no-cli-auto-prompt	boolean	Disable automatic prompt for CLI input parameters.
--ca-bundle <file>	string	The CA certificate bundle to use when verifying SSL certificates.
--endpoint-url <url>	string	Override the default API endpoint URL.
--no-verify-ssl	boolean	Do not verify SSL certificates.
--no-sign-request	boolean	Do not sign requests (for local providers).

Core Commands

ask — AI Conversations

Start conversations with AI providers. Supports streaming responses, file attachments, act templates, and session management.

Quick Examples:

# Simple question
rawi ask "Explain async/await in JavaScript"

# With file context
rawi ask "Review this code" --file src/app.js

# Using act template
rawi ask "Optimize this SQL" --act database_expert

# Continue previous session
rawi ask "Add error handling" --continue

Use Cases:

• Code reviews and debugging
• Technical explanations
• Creative writing
• Data analysis
• Problem solving

configure — Setup & Management

Configure AI providers, manage profiles, and customize Rawi’s behavior. Supports interactive setup and manual configuration.

Quick Examples:

# Interactive setup (recommended for first-time users)
rawi configure

# Manual OpenAI setup
rawi configure --provider openai --model gpt-4o --api-key sk-your-key

# Work profile setup
rawi configure --profile work --provider azure --resource-name company-openai

# List all profiles
rawi configure --list

Use Cases:

• Initial setup and onboarding
• Multiple provider management
• Team configuration standardization
• Environment-specific settings

history — Conversation Management

Search, manage, export, and analyze your AI conversation history and sessions.

Quick Examples:

# Show recent conversations
rawi history

# Search conversations
rawi history --search "docker deployment"

# Export history
rawi history export --format json --output backup.json

# Usage statistics
rawi history stats --from 2024-01-01

Use Cases:

• Finding previous solutions
• Conversation analytics
• Data export and backup
• Usage monitoring

info — System Information

Get detailed information about your Rawi installation, configuration, and system status.

Quick Examples:

# System overview
rawi info

# Profile details
rawi info --profiles

# Provider status
rawi info --providers

Use Cases:

• Configuration troubleshooting
• System health monitoring
• Support information gathering
• Setup verification

exec — Command Execution

Convert natural language descriptions into executable CLI commands with safety validation and confirmation prompts.

Quick Examples:

# Simple file operations
rawi exec "list all files in current directory"

# Preview dangerous operations
rawi exec "delete old log files" --dry-run

# System operations
rawi exec "check memory usage"

Use Cases:

• Server management and monitoring
• Complex file operations and data processing
• Development workflows and git operations
• Learning CLI tools and best practices

Supported AI Providers

Provider	Models Available	API Key Required	Local	Best For
OpenAI	GPT-4, GPT-4o, GPT-3.5	Yes	No	General, coding, creative
Anthropic	Claude 3.5 Sonnet	Yes	No	Safety, long-form analysis
Google	Gemini 1.5 Pro	Yes	No	Fast, multimodal
Ollama	Llama 2, CodeLlama	No	Yes	Privacy, offline, free
Azure	GPT-4, GPT-3.5	Yes + Resource	No	Enterprise, compliance
Bedrock	Claude, Llama	AWS Credentials	No	AWS integration, enterprise
xAI	Grok	Yes	No	Real-time, conversational
LM Studio	Various Local Models	No	Yes	Local hosting, custom models

Provider Setup Quick Reference

# OpenAI
rawi configure --provider openai --model gpt-4o --api-key sk-your-key

# Anthropic
rawi configure --provider anthropic --model claude-3-5-sonnet-20241022 --api-key sk-ant-your-key

# Ollama (local)
rawi configure --provider ollama --model llama2

# Azure OpenAI
rawi configure --provider azure --model gpt-4 --api-key your-key --resource-name your-resource

# AWS Bedrock
rawi configure --provider bedrock --model anthropic.claude-3-sonnet --region us-east-1 --use-provider-chain

Command Patterns

Basic Usage Patterns

# Question → Answer
rawi ask "What is Docker?"

# Configuration → Usage
rawi configure --provider openai
rawi ask "Hello world"

# History → Search → Reuse
rawi history --search "deployment"
rawi ask "Continue that deployment discussion" --continue abc123

Advanced Workflows

# Code Review Workflow
rawi ask "Review this PR" --file changes.diff --act code_reviewer --profile work

# Data Analysis Workflow
rawi ask "Analyze this dataset" --file data.csv --act data_analyst --temperature 0.1

# Writing Workflow
rawi ask "Write a blog post about AI" --act content_writer --temperature 0.8 --max-tokens 3000

Automation Patterns

# Automated code review
git diff HEAD~1 | rawi ask "Review these changes" --act code_reviewer --profile ci

# Daily standup prep
rawi ask "Summarize yesterday's commits" --file $(git log --since="1 day ago" --pretty=format:"%h %s")

# Documentation generation
rawi ask "Generate README for this project" --file package.json --act documentation_expert

Environment Integration

Shell Integration

Add to your .bashrc or .zshrc:

# Quick ask alias
alias ask='rawi ask'

# Code review alias
alias review='rawi ask --act code_reviewer'

# Git commit message helper
alias commit-msg='git diff --cached | rawi ask "Write a commit message for these changes" --max-tokens 100'

IDE Integration

Most IDEs can be configured to run Rawi commands:

# VS Code task
{
  "label": "AI Code Review",
  "type": "shell",
  "command": "rawi ask 'Review the current file' --file ${file} --act code_reviewer"
}

CI/CD Integration

# GitHub Actions example
- name: AI Code Review
  run: |
    git diff origin/main...HEAD | rawi ask "Review this PR" --act code_reviewer --profile ci > review.md
    gh pr comment --body-file review.md

Configuration Management

Profile Strategy

# Development profiles
rawi configure --profile dev-fast --provider openai --model gpt-3.5-turbo --temperature 0.3
rawi configure --profile dev-thorough --provider anthropic --model claude-3-5-sonnet --temperature 0.1

# Work profiles
rawi configure --profile work --provider azure --model gpt-4 --resource-name company-ai
rawi configure --profile personal --provider openai --model gpt-4o --api-key personal-key

# Specialized profiles
rawi configure --profile code-review --provider anthropic --temperature 0.1 --max-tokens 4000
rawi configure --profile creative --provider openai --temperature 0.9 --max-tokens 3000

Environment Variables

# Default configuration
export RAWI_PROFILE="work"
export RAWI_PROVIDER="openai"
export RAWI_MODEL="gpt-4"

# API keys
export OPENAI_API_KEY="sk-your-openai-key"
export ANTHROPIC_API_KEY="sk-ant-your-anthropic-key"
export GOOGLE_API_KEY="your-google-key"

Best Practices

Security

1. API Key Management: Use environment variables for production
2. Profile Separation: Different profiles for different security contexts
3. Local Providers: Use Ollama for sensitive data
4. History Management: Regular cleanup of conversation history

Performance

1. Model Selection: Choose appropriate models for your use case
2. Token Limits: Set reasonable limits to control costs
3. Temperature Settings: Lower for factual tasks, higher for creative
4. Streaming: Enable streaming for real-time feedback

Organization

1. Meaningful Profiles: Use descriptive names like work-gpt4, local-dev
2. Consistent Naming: Standardize profile naming across teams
3. Documentation: Document your profile configurations
4. Regular Maintenance: Clean up unused profiles and history

Troubleshooting

Common Issues

Command not found:

# Check installation
which rawi
# Reinstall if needed
npm install -g rawi

Configuration errors:

# Check configuration
rawi info --profiles
# Reset if needed
rawi configure --reset

Provider connection issues:

# Test configuration
rawi configure --test
# Check network connectivity
ping api.openai.com

History issues:

# Check history location
rawi info | grep "History:"
# Fix permissions if needed
chmod 755 ~/.config/rawi/

Getting Help

# Command help
rawi <command> --help

# System information
rawi info

# Debug mode
rawi --debug ask "test question"

# Verbose output
rawi ask "test" --verbose