CLI Reference¶

Oblix provides a powerful command-line interface (CLI) that enables you to interact with the Oblix AI Orchestration SDK directly from your terminal. This tool simplifies hybrid model orchestration between cloud and edge environments.

Installation¶

The Oblix CLI is included with the main Oblix SDK package when you install it via pip:

pip install oblix

Note: Currently, Oblix only supports macOS. Windows and Linux versions are planned for future releases. Supported Python versions: 3.9, 3.10, 3.11, 3.12, 3.13.

Basic Usage¶

Once installed, you can access the CLI by using the oblix command in your terminal:

oblix --help

This will display the available commands and options.

Command Structure¶

The Oblix CLI follows this general structure:

oblix [command] [subcommand] [options]

Orchestration Capabilities¶

Oblix is designed specifically for orchestrating AI workloads between cloud and edge environments. Rather than running models in isolation, Oblix's CLI provides:

Automatic orchestration between local and cloud models based on resource availability and connectivity
Dynamic routing of prompts to the most appropriate model
Monitoring agents that continuously assess system state
Seamless fallback between models when conditions change

Available Commands¶

Command	Description
`models`	Manage and interact with AI models (both local and cloud)
`agents`	Manage and interact with Oblix orchestration agents
`sessions`	Manage chat sessions
`execute`	Execute a single prompt with orchestration
`chat`	Start an interactive chat session with hybrid model orchestration

Authentication¶

Most commands require an Oblix API key, which can be provided in several ways:

Using the --api-key option with each command
Setting the OBLIX_API_KEY environment variable
When not provided, the CLI will prompt you to enter it

Example:

export OBLIX_API_KEY="your_oblix_api_key"

Common Options¶

These options are available for most commands:

Option	Description
`--help`, `-h`	Display help for a command
`--version`	Show the Oblix CLI version
`--debug`	Enable debug logging

Models Commands¶

Commands for managing AI models across cloud and edge:

List Models¶

List all configured models:

oblix models list

Hook Model¶

Add a new model to the orchestration system:

# Hook a local Ollama model
oblix models hook --type ollama --name llama2

# Hook a cloud OpenAI model
oblix models hook --type openai --name gpt-3.5-turbo --api-key-model YOUR_OPENAI_KEY

# Hook a cloud Claude model
oblix models hook --type claude --name claude-3-haiku-20240307 --api-key-model YOUR_ANTHROPIC_KEY

Model Info¶

Get information about a specific model:

oblix models info --type ollama --name llama2

Agents Commands¶

Commands for interacting with Oblix orchestration agents:

Connectivity¶

Check current connectivity metrics (critical for orchestration decisions):

oblix agents connectivity

Resource¶

Perform a system resource check (impacts model routing decisions):

oblix agents resource

Sessions Commands¶

Commands for managing orchestrated chat sessions:

List Sessions¶

List recent chat sessions:

oblix sessions list

View Session¶

View details of a specific session:

oblix sessions view SESSION_ID

Delete Session¶

Delete a specific chat session:

oblix sessions delete SESSION_ID

Create Session¶

Create a new chat session:

oblix sessions create --title "My New Chat"

Orchestrated Execution and Chat¶

Execute prompts with automatic orchestration between cloud and edge:

Execute a Prompt with Orchestration¶

Let Oblix decide which model to use based on system state:

oblix execute "Explain quantum computing in simple terms"

With a specific model (bypassing orchestration):

oblix execute "Explain quantum computing" --model "openai:gpt-3.5-turbo"

Start Interactive Chat with Hybrid Orchestration¶

Start an interactive chat session with automatic orchestration between cloud and edge models:

oblix chat --local-model llama2 --cloud-model gpt-3.5-turbo

With Claude as the cloud model:

oblix chat --local-model llama2 --cloud-model claude-3-haiku --cloud-api-key YOUR_ANTHROPIC_KEY

Orchestration Examples¶

Here are some common orchestration use cases for the Oblix CLI:

Setting Up Hybrid Model Orchestration¶

# Set up local and cloud models for orchestration
oblix models hook --type ollama --name llama2
oblix models hook --type openai --name gpt-3.5-turbo --api-key-model YOUR_OPENAI_KEY

# Start interactive chat with model orchestration
oblix chat --local-model llama2 --cloud-model gpt-3.5-turbo

Checking Orchestration System Status¶

# Check system resource metrics (affects local model selection)
oblix agents resource

# Check connectivity metrics (affects cloud model selection)
oblix agents connectivity

Managing Orchestrated Sessions¶

# List recent sessions
oblix sessions list

# Create a new session
oblix sessions create --title "Research Project"

# Execute with a specific session
oblix execute "Research quantum computing papers" --session-id YOUR_SESSION_ID

Environment Variables¶

Variable	Description
`OBLIX_API_KEY`	Your Oblix API key
`OPENAI_API_KEY`	Your OpenAI API key (used as default for OpenAI models)
`ANTHROPIC_API_KEY`	Your Anthropic API key (used as default for Claude models)

Troubleshooting¶

If you encounter issues with the CLI:

Ensure you're using the latest version of Oblix
Check that you've properly set up your API keys
Enable debug logging with the --debug flag
Verify that the Ollama server is running if using local models
Confirm your Mac meets the system requirements (particularly for larger models)
Check network connectivity if cloud models aren't responding

For additional help, join the Oblix Discord community at https://discord.gg/v8qtEVuU.