Skip to main content
The AX CLI (ax) lets you manage your Arize AI platform resources — datasets, projects, spans, and configuration profiles — directly from your terminal. It’s useful for scripting workflows, automating data pipelines, and integrating Arize into CI/CD environments.

​
Getting Started

​
1. Install the CLI

Requires Python 3.11 or later.
pip install arize-ax-cli
Verify the installation: 
ax --version

​
2. Create a configuration profile

ax profiles create
You’ll be prompted for your API key, region, and preferred output format. Choose Simple mode for most use cases, or Advanced mode for custom routing, transport tuning, and SSL settings.
ax profiles create is required before running any other commands. Without it, most commands will fail with: Error: Profile 'default' not found.

​
3. Verify your setup

ax profiles show    # Check your profile settings
ax projects list    # Run your first command

​
Shell Completion (optional)

Enable tab completion for your shell: 
ax --install-completion
Supports bash, zsh, fish, and PowerShell. Restart your shell or source your profile to activate.

​
Configuration

Profiles store your API key, region, and output preferences so you don’t have to pass them on every command. Configuration files are stored at:
  • macOS/Linux: ~/.arize/profiles/<profile>.toml
  • Windows: %USERPROFILE%\.arize\profiles\<profile>.toml
The CLI automatically detects any ARIZE_* environment variables in your shell and offers to use them during ax profiles create. You can also reference environment variables in config files using ${VARIABLE_NAME} syntax.

​
Managing Profiles

Here’s a full reference of profile commands: 
ax profiles create                       # Create a new profile
ax profiles list                         # List all profiles (active profile is marked)
ax profiles show                         # Show active profile settings
ax profiles show --profile myprofile     # Show a specific profile
ax profiles show --expand                # Resolve environment variable references
ax profiles show --all                   # Show all sections including defaults
ax profiles use <profile>                # Switch the active profile
ax profiles delete <profile>             # Delete a profile (cannot delete default or active)
ax profiles delete <profile> --force     # Delete without confirmation prompt

​
Working with Multiple Environments

Create separate profiles for production and staging, then switch between them as needed: 
ax profiles create   # name it "production"
ax profiles create   # name it "staging"

# Switch the active profile
ax profiles use production

# Or override per-command with --profile
ax datasets list --profile staging --space-id sp_stg123

​
Options & Output

​
Top-Level Options

These options are available on the root ax command itself:
OptionShortDescription
--versionShow version and exit
--install-completionInstall shell completion
--show-completionShow completion script for your shell
--help-hShow help

​
Subcommand Options

Most subcommands (e.g., ax datasets list, ax projects get) accept these options:
OptionShortDescription
--profile <name>-pUse a specific configuration profile instead of the active one
--output <format|path>-oSet output format (table, json, csv, parquet) or write to a file path
--verbose-vEnable detailed logging
--help-hShow help for the command
--profile and --output are subcommand options, not top-level flags. Use them after the subcommand: ax datasets list --profile staging, not ax --profile staging datasets list.

​
Output Formats

The default output format is table. Use --output to change the format or write results directly to a file:
FormatDescription
tableHuman-readable default; formatted for the terminal
jsonMachine-parseable JSON
csvSpreadsheet-compatible format
parquetApache columnar format for large datasets
ax datasets list --output json                                      # Print as JSON
ax datasets list --output datasets.json                             # Write to a JSON file
ax datasets list_examples <dataset-id> --output examples.parquet    # Write to Parquet

​
Cache

The CLI caches certain data locally at ~/.arize/cache/ to speed up repeated operations. To clear the cache: 
ax cache clear

​
Resources

Datasets

Create, list, and manage datasets

Projects

Create, list, and manage projects

Spans

Query and export span data

Experiments

Create, list, and manage experiments

Traces

Query and export trace data

PyPI Package

Install and version history

GitHub Repository

Source code and issues

Arize Community Slack

Get help from the community

Support

Contact the Arize support team