Pollaris Docs
Pollaris CLI

Install the CLI

Install Pollaris, verify the connection, and prepare it for an AI agent.

Pollaris CLI requires Python 3.11 or newer. Install it in an isolated environment so its dependencies do not alter your project.

The CLI is available on PyPI.

Install

With uv:

uv tool install pollaris-cli

Or with pipx:

pipx install pollaris-cli

Confirm the installed command:

pollaris --version
export POLLARIS_API_URL="https://api.pollaris.cc"
pollaris capabilities --json

Upgrade

uv tool upgrade pollaris-cli

For a local MCP bridge, install the optional dependencies:

uv tool install 'pollaris-cli[mcp]'

API configuration

Use https://api.pollaris.cc as the production API URL, without /v1/. The CLI appends /v1/agent to its requests. Set POLLARIS_API_URL in your agent's environment so later invocations use the same deployment.

For local or staging deployments, set that variable to the relevant API origin. Pollaris accepts plain HTTP only for loopback development addresses such as localhost. Use HTTPS for remote API deployments.

Credential storage

The CLI stores access credentials in a supported operating-system keychain. It never prints credentials in normal or JSON output.

Headless environments without a keychain must supply an explicit Fernet key through their secret manager:

export POLLARIS_CLI_SECRET_KEY="YOUR_FERNET_KEY"

Do not commit that value or paste it into a prompt. If neither secure option is available, the CLI stops instead of saving credentials as plain text.

Start authorization manually

You normally do not need this step because a create command begins authorization automatically. To connect first:

pollaris auth login --json

Open the returned authorization.url. After approval, resume with the exact next_action.command returned by Pollaris.

On this page