ados 0.1.3

AI-powered coding agent for your terminal - like Claude Code

AutoDevOS CLI

AI-powered coding assistant for your terminal — like Claude Code.

Installation

Quick Install (Recommended)

# Using pip
pip install ados

# Or using pipx (isolated environment)
pipx install ados

From Source

git clone https://github.com/KhalkarYash/autodevos-cli.git
cd autodevos-cli
pip install -e .

Quick Start

1. Setup (First Time)

# Interactive setup wizard
ados auth setup

This will guide you through:

• Choosing your LLM provider (OpenAI, Claude, Gemini, Ollama, etc.)
• Setting up your API key
• Configuring your preferred model

2. Start Coding

# Open in current directory
ados

# Open a specific project
ados ~/projects/myapp

# Run a single command
ados "create a simple todo app with React"

Supported Providers

Provider	API Key	Default Model
ollama	Not needed	llama3.2
openai	Required	gpt-4o
anthropic	Required	claude-sonnet-4-20250514
gemini	Required	gemini-1.5-pro
openrouter	Required	anthropic/claude-sonnet-4-20250514

Commands

Main Commands

ados                    # Interactive session in current directory
ados <directory>        # Open specific directory
ados "prompt"           # Single prompt mode
ados --init             # Initialize project config

Configuration

ados config             # Show current config
ados config provider    # List/set provider
ados config model       # Set model
ados config url         # Set custom API URL
ados config providers   # List all providers

Authentication

ados auth               # Show auth status
ados auth setup         # Interactive setup wizard
ados auth set-key <provider>  # Set API key
ados auth remove-key <provider>  # Remove API key

Session Commands (Interactive Mode)

Command	Description
/help	Show available commands
/config	Show current configuration
/tools	List available tools
/stats	Show session statistics
/save	Save current session
/clear	Clear conversation
/exit	Exit the session

Configuration

Global Config

Located at ~/.config/autodevos/config.toml:

[model]
provider = "anthropic"
name = "claude-sonnet-4-20250514"
temperature = 1.0

Project Config

Create .autodevos/config.toml in your project:

# Project-specific AI instructions
developer_instructions = """
This is a Python FastAPI project.
Follow PEP 8 style guidelines.
Use pytest for testing.
"""

# Tool restrictions
allowed_tools = ["read_file", "write_file", "shell", "grep"]

# Approval policy
approval = "auto"

Team Setup (Dashboard Integration)

For teams, admins can generate a setup token:

# Admin generates token
ados auth generate-token
# Output: eyJwcm92aWRlci...

# Team members apply it
ados auth setup-token eyJwcm92aWRlci...
ados auth set-key anthropic  # Set their own API key

Using with Ollama (Local)

# Start Ollama
ollama serve

# Configure AutoDevOS
ados config provider ollama
ados config model llama3.2

# Start coding!
ados

Remote Ollama (Different Machine)

# On the machine running Ollama:
OLLAMA_HOST=0.0.0.0 ollama serve

# On your machine:
ados config provider ollama
ados config url http://192.168.1.100:11434/v1

Features

• Multi-Provider Support: OpenAI, Anthropic, Google, Ollama, and more
• Interactive Terminal UI: Rich formatting with streaming responses
• Built-in Tools: File operations, shell commands, web search, and more
• Context Management: Automatic compression for long conversations
• Session Persistence: Save and resume conversations
• Safety Controls: Approval policies for dangerous operations
• MCP Integration: Extend with Model Context Protocol servers
• Project Config: Customize AI behavior per-project

Environment Variables

Override config with environment variables:

export API_KEY="your-api-key"
export BASE_URL="https://custom-api.example.com/v1"

Requirements

• Python 3.11+
• For Ollama: Ollama installed and running

License

MIT