VM0 CLI

Build and run AI agents in secure sandboxed environments using the VM0 command-line interface.

Official docs: https://docs.vm0.ai

When to Use

Use this skill when you need to:

Install and set up the VM0 CLI Create and configure AI agent projects Deploy agents to the VM0 platform Run agents with prompts and inputs Manage input files (volumes) and output files (artifacts) View logs and usage statistics Prerequisites Installation

Install the VM0 CLI globally via npm:

npm install -g @vm0/cli

Verify installation:

vm0 --version

Authentication

vm0 auth login

This opens a browser for authentication. After login, verify status:

vm0 auth status

For CI/CD environments, get your API token:

vm0 auth setup-token

Then set the environment variable:

export VM0_TOKEN=vm0_live_your-api-key

Quick Start 1. Initialize a Project

Create a new VM0 project in the current directory:

vm0 init

This creates a vm0.yaml configuration file interactively. For non-interactive mode:

vm0 init --name my-agent

Configure the Agent

Edit vm0.yaml to define your agent:

version: "1.0"

agents: my-agent: framework: claude-code instructions: AGENTS.md skills: - https://github.com/vm0-ai/vm0-skills/tree/main/github environment: DEBUG: "${{ vars.DEBUG }}" API_KEY: "${{ secrets.API_KEY }}"

Deploy the Agent

Deploy your agent configuration:

vm0 compose vm0.yaml

Skip confirmation prompts with -y:

vm0 compose vm0.yaml -y

Run the Agent

Execute the agent with a prompt:

vm0 run my-agent "Please analyze the codebase and suggest improvements"

Or use cook for one-click execution from vm0.yaml:

vm0 cook "Analyze the code"

Core Operations Running Agents

Basic run:

vm0 run my-agent "Your prompt here"

Run with variables and secrets:

vm0 run my-agent "Process data" --vars DEBUG=true --secrets API_KEY=xxx

Run with artifact storage:

vm0 run my-agent "Generate report" --artifact-name my-output

Run with input volumes:

vm0 run my-agent "Process files" --volume-version input-data=latest

Enable verbose output:

vm0 run my-agent "Hello" -v

Resume from checkpoint:

vm0 run resume "Continue the task"

Continue from session:

vm0 run continue "Next step"

One-Click Execution (cook)

Run directly from vm0.yaml in current directory:

vm0 cook "Your prompt"

Skip confirmation:

vm0 cook -y "Your prompt"

Continue last session:

vm0 cook continue "Follow up"

Resume from last checkpoint:

vm0 cook resume "Continue"

View logs from last cook run:

vm0 cook logs

Viewing Logs

View agent events (default):

vm0 logs

View system logs:

vm0 logs --system

View metrics:

vm0 logs --metrics

View network logs:

vm0 logs --network

Filter by time:

vm0 logs --since 5m vm0 logs --since 2h vm0 logs --since 2024-01-15T10:30:00Z

Show last N entries:

vm0 logs --tail 20

Storage Management Volumes (Input Files)

Volumes store input files that agents can read.

Initialize a volume (interactive):

cd my-data-directory vm0 volume init

Initialize a volume (non-interactive):

cd my-data-directory vm0 volume init --name my-data

Push local files to cloud:

vm0 volume push

Pull cloud files to local:

vm0 volume pull

Pull specific version:

vm0 volume pull abc123de

Check volume status:

vm0 volume status

List all volumes:

vm0 volume list

Clone a remote volume:

vm0 volume clone my-volume ./local-dir

Artifacts (Output Files)

Artifacts store output files created by agents.

Initialize an artifact (interactive):

cd my-output-directory vm0 artifact init

Initialize an artifact (non-interactive):

cd my-output-directory vm0 artifact init --name my-output

Push local files to cloud:

vm0 artifact push

Pull cloud files to local:

vm0 artifact pull

Pull specific version:

vm0 artifact pull abc123de

Check artifact status:

vm0 artifact status

List all artifacts:

vm0 artifact list

Clone a remote artifact:

vm0 artifact clone my-artifact ./local-dir

Agent Management List Agents vm0 agent list

With details:

vm0 agent list --verbose

Inspect Agent

View agent configuration:

vm0 agent inspect my-agent

View specific version:

vm0 agent inspect my-agent:abc123

Usage Statistics

View your usage statistics:

vm0 usage

Filter by date range:

vm0 usage --since 7d vm0 usage --since 30d vm0 usage --since 2024-01-01 --until 2024-01-31

Model Provider Configuration

Manage LLM model providers for agent runs.

List providers:

vm0 model-provider list

Setup a provider (interactive):

vm0 model-provider setup

Setup a provider (non-interactive):

vm0 model-provider setup --type anthropic-api-key --credential "sk-ant-xxx"

Set default provider:

vm0 model-provider set-default anthropic-api-key

Delete a provider:

vm0 model-provider delete anthropic-api-key

Credential Management

Store credentials for agent runs.

List credentials:

vm0 credential list

Set a credential:

vm0 credential set MY_API_KEY "secret-value"

Delete a credential:

vm0 credential delete MY_API_KEY

vm0.yaml Reference Basic Structure version: "1.0"

agents: agent-name: framework: claude-code instructions: AGENTS.md skills: - https://github.com/vm0-ai/vm0-skills/tree/main/github - https://github.com/vm0-ai/vm0-skills/tree/main/slack environment: VAR_NAME: "value" SECRET_VAR: "${{ secrets.SECRET_NAME }}" CONFIG_VAR: "${{ vars.CONFIG_NAME }}"

Fields Field Description version Configuration version (currently "1.0") agents Map of agent definitions framework Agent framework (e.g., claude-code) instructions Path to instructions file skills List of skill URLs to include environment Environment variables for the agent Variable Syntax ${{ secrets.NAME }} - Sensitive values stored securely ${{ vars.NAME }} - Non-sensitive configuration values Direct values - Plain text values Environment Information

View system and environment details:

vm0 info

Non-Interactive Mode (CI/CD)

All commands support non-interactive mode for use in CI/CD pipelines, scripts, and automated environments. The CLI detects non-TTY environments (process.stdout.isTTY === false) and requires explicit flags for all inputs.

Authentication

Set the VM0_TOKEN environment variable instead of interactive login:

export VM0_TOKEN=vm0_live_your-api-key

Command Reference Command Non-Interactive Flags Notes vm0 init --name Required in non-TTY vm0 compose -y, --yes Skip new secrets confirmation vm0 cook -y, --yes Skip all confirmation prompts vm0 volume init --name Required in non-TTY vm0 artifact init --name Required in non-TTY vm0 model-provider setup --type --credential Both required together CI/CD Example

Set authentication

export VM0_TOKEN=${{ secrets.VM0_TOKEN }}

Initialize project (non-interactive)

vm0 init --name my-agent --force

Initialize storage (non-interactive)

cd input-data && vm0 volume init --name input-data && cd .. cd artifact && vm0 artifact init --name artifact && cd ..

Deploy agent (skip confirmation)

vm0 compose vm0.yaml -y

Run agent

vm0 run my-agent --artifact-name artifact "Execute the task"

GitHub Actions Example jobs: run-agent: runs-on: ubuntu-latest steps: - uses: actions/checkout@v4

  - name: Setup Node.js
    uses: actions/setup-node@v4
    with:
      node-version: '20'

  - name: Install VM0 CLI
    run: npm install -g @vm0/cli

  - name: Run Agent
    env:
      VM0_TOKEN: ${{ secrets.VM0_TOKEN }}
    run: |
      vm0 compose vm0.yaml -y
      vm0 run my-agent --artifact-name output "Generate daily report"

Model Provider Setup (Non-Interactive) vm0 model-provider setup --type anthropic-api-key --credential "sk-ant-xxx"

Guidelines Always authenticate first - Run vm0 auth login before using other commands Use vm0 init for new projects - Creates proper project structure Deploy before running - Run vm0 compose after modifying vm0.yaml Use volumes for input data - Push data files as volumes before running agents Check logs for debugging - Use vm0 logs to troubleshoot failed runs Use scopes for organization - Set appropriate scope for team collaboration Common Workflows Deploy and Run Agent

1. Initialize project

vm0 init --name my-agent

2. Edit vm0.yaml and AGENTS.md

3. Deploy configuration

vm0 compose vm0.yaml

4. Run the agent

vm0 run my-agent "Execute the task"

5. Check logs if needed

vm0 logs

Provide Input Files to Agent

1. Create and navigate to data directory

mkdir input-data && cd input-data

2. Add your files

cp ~/documents/*.pdf .

3. Initialize and push volume (use --name for non-interactive)

vm0 volume init --name input-data vm0 volume push

4. Run agent with volume

cd .. vm0 run my-agent "Process the documents" --volume-version input-data=latest

Download Agent Output

1. List artifacts

vm0 artifact list

2. Clone the artifact locally

vm0 artifact clone my-output ./results

3. Or pull to existing directory

cd my-output-dir vm0 artifact pull

Troubleshooting Authentication Issues

Check auth status

vm0 auth status

vm0 auth logout vm0 auth login

Agent Not Found

List available agents

vm0 agent list

Check if deployed

vm0 compose vm0.yaml

View Detailed Errors

Use verbose mode

vm0 run my-agent "prompt" -v

Check system logs

vm0 logs --system

vm0-cli

安装

Set authentication

Initialize project (non-interactive)

Initialize storage (non-interactive)

Deploy agent (skip confirmation)

Run agent

1. Initialize project

2. Edit vm0.yaml and AGENTS.md

3. Deploy configuration

4. Run the agent

5. Check logs if needed

1. Create and navigate to data directory

2. Add your files

3. Initialize and push volume (use --name for non-interactive)

4. Run agent with volume

1. List artifacts

2. Clone the artifact locally

3. Or pull to existing directory

Check auth status

List available agents

Check if deployed

Use verbose mode

Check system logs

安装

Set authentication

Initialize project (non-interactive)

Initialize storage (non-interactive)

Deploy agent (skip confirmation)

Run agent

1. Initialize project

2. Edit vm0.yaml and AGENTS.md

3. Deploy configuration

4. Run the agent

5. Check logs if needed

1. Create and navigate to data directory

2. Add your files

3. Initialize and push volume (use --name for non-interactive)

4. Run agent with volume

1. List artifacts

2. Clone the artifact locally

3. Or pull to existing directory

Check auth status

Re-login if needed

List available agents

Check if deployed

Use verbose mode

Check system logs