Anaconda CLI

The command-line interface for the Anaconda platform.

Table of contents

• When to use this
• Project status
• Prerequisites
• Installation
• Quick start
• Common workflows
• Configuration
• Command reference
• Troubleshooting
• Documentation
• License

When to use this

The Anaconda CLI is the entry point for the Anaconda ecosystem. It handles authentication to Anaconda services, configures tools like conda and pixi, and manages access to package channels, so you don't have to manually edit config files or manage credentials across tools separately.

If you're building with Anaconda's data science, ML, or AI tooling, ana gets you from zero to a working environment in minutes.

Project status

Anaconda CLI is under active development and follows semantic versioning. Commands not marked as experimental are considered stable. Features behind ana feature enable may change between releases.

Prerequisites

• macOS (Apple Silicon), Linux (x86_64 or aarch64), or Windows (x86_64)
• conda installed if you plan to create conda environments
• Internet connection required on first run on macOS (Gatekeeper notarization check)

Installation

macOS and Linux

curl -fsSL https://anaconda.sh/install.sh | sh

Windows (PowerShell)

powershell -ExecutionPolicy ByPass -c "irm https://anaconda.sh/install.ps1 | iex"

Installation options

Bash	PowerShell	Environment variable	Default	Description
-d, --install-dir	-InstallDir	ANA_INSTALL_DIR	~/.local/bin	Installation directory
-v, --version	-Version	ANA_VERSION	latest	Version to install
--no-verify-checksum	-NoVerifyChecksum	ANA_VERIFY_CHECKSUM	true	Set to false to skip checksum validation
--no-path-update	-NoPathUpdate	ANA_NO_PATH_UPDATE		Set to skip shell profile modification
-f, --force	-Force	ANA_FORCE_INSTALL		Set to overwrite without prompting

Quick start

# 1. Verify installation
ana --version

# 2. Log in to your Anaconda account (opens browser if available)
ana login

# 3. Enable early-access packages from the main-x channel
ana feature enable main-x

# 4. Create your first environment
conda create -n myproject python=3.12 numpy pandas jupyter
conda activate myproject

Common workflows

Configuring conda and pixi for Anaconda channels

The CLI can configure both conda and pixi to use Anaconda channels. Install pixi separately if you prefer it over conda, then point it at main-x:

ana feature enable main-x

CI/CD authentication

For non-interactive environments, authenticate with an API key via stdin:

echo "$ANACONDA_API_KEY" | ana login --api-key

Integrating AI assistants with your environment

ana provides MCP (Model Context Protocol) support so AI assistants like Claude can discover and work with packages in your conda environments:

ana mcp setup
ana mcp discover

See ana mcp --help for the full list of subcommands.

Deploying ML workflows to Outerbounds (experimental)

An experimental integration with the Outerbounds platform is available for production ML workflows. See the Outerbounds integration guide for setup and usage.

ana feature enable outerbounds
ana ob configure --instance your-org.outerbounds.com
ana ob init my-pipeline --name recommendation-engine
ana ob deploy

Configuration

View current settings with ana config.

Environment variables

Variable	Default	Description
ANA_DOMAIN	anaconda.com	Authentication domain
ANA_SSL_VERIFY	true	Verify SSL certificates
ANA_OPEN_BROWSER	true	Auto-open browser during login
ANA_ENABLE_TELEMETRY	true	Send anonymous usage telemetry (see telemetry policy)
ANA_PRERELEASES	false	Include pre-releases in update checks
ANA_KEYRING_PATH	~/.anaconda/keyring	API key storage location

Boolean values accept true/false, 1/0, or empty string (treated as false).

Feature flags

Experimental features are stored in ~/.ana/config.toml:

[ana.features]
outerbounds = true

Command reference

Authentication

ana login                          # Interactive login (browser-based device flow)
ana login --api-key "your-key"     # Login with API key directly
echo "$KEY" | ana login --api-key  # Login with API key from stdin
ana whoami                         # View account info
ana whoami --json                  # Account info as JSON
ana logout                         # Log out

### Tool management

```bash
ana tool list                    # List available tools
ana tool install <tool>          # Install a tool
ana tool uninstall <tool>        # Uninstall a tool

Package features

ana feature enable main-x        # Enable early-access packages (conda and pixi)
ana feature enable main-x --pixi # Enable for pixi only
ana feature disable main-x       # Disable a feature

Self-management

ana self update                  # Update to latest version
ana self update v1.0.0           # Update to specific version
ana self update --check          # Check for updates without installing
ana self update --list           # List available versions

MCP (Model Context Protocol)

ana mcp setup                    # Configure AI clients for Anaconda MCP
ana mcp discover                 # Discover MCP servers from environments
ana mcp clients                  # List supported AI clients
ana mcp serve                    # Start MCP servers
ana mcp remove                   # Remove MCP configuration

API access

ana api fetch /api/account                                        # GET request
ana api fetch /api/endpoint --method POST --json '{"key": "val"}' # POST with JSON
ana api fetch /api/search -q "name=numpy,version=1.24"            # Query parameters

Anaconda CLI (classic)

If you need the classic anaconda command (for uploading packages to anaconda.org, etc.):

ana bootstrap

This installs the anaconda command, which you can then use directly:

anaconda upload my-package.tar.bz2
anaconda search numpy

Troubleshooting

"Command not found" after installation

The installer adds ~/.local/bin to your PATH via your shell profile. Restart your shell, or run:

export PATH="$HOME/.local/bin:$PATH"

Login fails with "invalid API key"

Verify your API key at anaconda.com. Keys are environment-specific (anaconda.com vs. self-hosted) and won't work cross-environment.

SSL certificate errors

If you're behind a corporate proxy or using custom CA certificates:

export ANA_SSL_VERIFY=false
ana login

Only disable SSL verification in trusted network environments.

Verbose logging

ana -vvv login                   # Up to 5 verbosity levels
RUST_LOG=ana=debug ana login      # Fine-grained control via RUST_LOG

Documentation

• Code of Conduct
• Contributing guide

License

Anaconda CLI is licensed under the Apache License 2.0. See LICENSE for details.