CLI Reference

The tensorify CLI authenticates with Tensorify, runs a runner on your machine, and exports workflow code.

Linux & macOS:

curl -fsSL https://cli.tensorify.io/install | sh

Windows (PowerShell):

irm https://cli.tensorify.io/install.ps1 | iex

tensorify --version

Local machine (has a browser):

tensorify init
tensorify login
tensorify runner start

VPS / headless server (no browser):

tensorify init --name my-vps --api-key tfk_your_key_here
tensorify runner start

Docker container:

docker run -e TENSORIFY_API_KEY=tfk_your_key my-runner

Get an API key from Settings → API Keys in the web app. On a VPS, use sudo tensorify runner install to run the runner as a systemd service instead of keeping a terminal open. See Installation for Docker image setup.

Create the local runner configuration file.

tensorify init [options]

Creates ~/.tensorify/config.json with your runner name, API endpoints, and optionally an API key. If --name is not passed and stdin is a terminal, you will be prompted interactively. On non-interactive shells (e.g. SSH scripts), it defaults to the hostname.

# Local machine (browser login):
tensorify init
tensorify login
tensorify runner start

# VPS (API key, no browser needed):
tensorify init --name production-vps --api-key tfk_abc123
tensorify runner start

Start the multi-workflow runner in the foreground. One runner process can serve every workflow you deploy to it from the UI.

tensorify runner start [options]

Registers your machine with Tensorify and maintains a WebSocket connection. While connected:

• Test runs from the canvas editor execute on your machine
• Deployed workflows with CLI or Auto execution mode route real webhook and API triggers to this runner

Leave this process running during development. In production, use tensorify runner install or a process manager.

tensorify runner start

Install the runner as a system service (systemd on Linux, launchd on macOS).

sudo tensorify runner install    # Linux
tensorify runner install          # macOS (user-level launchd)

Creates a service unit that starts tensorify runner start on boot and restarts it on failure. On Linux, it also runs systemctl daemon-reload, enables the service, and starts it immediately. Use this on VPS and production servers instead of keeping a terminal open.

# Full VPS setup (two commands):
tensorify init --name production-vps --api-key tfk_your_key_here
sudo tensorify runner install

# Verify:
systemctl status tensorify-runner

Remove the systemd or launchd service installed by runner install.

sudo tensorify runner uninstall    # Linux
tensorify runner uninstall          # macOS

Authenticate the CLI with your Tensorify workspace.

tensorify login [options]

Without --api-key, opens a browser-based sign-in flow and stores your session locally. With --api-key, stores the key directly in config — no browser needed.

# Browser login (desktop):
tensorify login

# Headless login (VPS):
tensorify login --api-key tfk_abc123

Remove stored credentials from this machine.

tensorify logout

Show the current authenticated user and workspace.

tensorify whoami

Export a workflow as Python source code.

tensorify export <workflowId> [options]

Downloads the compiled Python files and a requirements.txt for the workflow. The output can be run as a standalone process or built into a Docker image without the Tensorify CLI.

tensorify export wf_abc123 --output ./my-workflow

cd my-workflow
pip install -r requirements.txt
python main.py

Show cached Python runtime versions stored locally.

tensorify runtime list

Lists all runtime wheel versions downloaded to ~/.tensorify/runtimes/. The CLI downloads runtime packages from runtime.tensorify.io on first use and caches them locally.

Remove all cached runtimes and virtual environments.

tensorify runtime clear

Deletes ~/.tensorify/runtimes/ and ~/.tensorify/cache/. The next workflow execution will re-download the runtime and re-create virtual environments. Use this to free disk space or resolve corrupted environments.

Update the CLI binary to the latest version.

tensorify self-update

Checks cli.tensorify.io/releases/latest.txt for the newest version. If a newer version is available, downloads the binary for your platform, verifies its SHA256 checksum, and replaces the current binary in-place.

These flags are available on all commands:

The CLI stores its configuration at ~/.tensorify/config.json:

{
  "runnerName": "my-server",
  "teamspaceId": "...",
  "apiKey": "tfk_...",
  "apiUrl": "https://app.tensorify.io",
  "hooksUrl": "https://triggers.tensorify.io",
  "cdnUrl": "https://runtime.tensorify.io"
}

This file is created by tensorify init and updated by tensorify login. Environment variables override values from this file.

• Running Workflows — runner setup, testing from the canvas, and production execution
• Deploying Workflows — configure execution modes and triggers from the UI