---
title: "Install Anyscale agent skills"
description: "Prerequisites and installation instructions for Anyscale agent skills, including how to list, install, and update skills using the Anyscale CLI."
---

# Install Anyscale agent skills

This page covers the prerequisites for Anyscale agent skills and how to install and update them using the Anyscale CLI.

## Prerequisites

Before using Anyscale agent skills, you need the following:

-   A supported AI coding agent: [Claude Code](https://docs.anthropic.com/en/docs/claude-code) or [Cursor](https://www.cursor.com/).
-   The Anyscale CLI installed and authenticated. See [CLI configuration](/reference/quickstart-cli.md).

Install or update the Anyscale CLI with pip, then log in:

```bash
pip install -U anyscale
anyscale login
```

Your use of Anyscale agent skills is governed by the [Anyscale agent skills terms and conditions](/agent-skills-terms.md).

## Manage skills

Install skills through the Anyscale CLI. The Anyscale CLI distributes skills from a private repository. You can't install Anyscale agent skills through plugins, a marketplace, or `npx`.

### List available skills

To see all available skills, run the following command:

```bash
# List all skills that you can install.
anyscale skills list
```

### Install skills

To install skills for your specific platform, run one of the following commands:

```bash
# Install skills for Claude Code.
anyscale skills install --platform claude-code

# Install skills for Cursor.
anyscale skills install --platform cursor

# Install skills for all supported platforms.
anyscale skills install --platform all
```

### Update skills

To keep your skills up to date, run the following command:

```bash
# Update all installed skills to their latest versions.
anyscale skills update
```

## Security acknowledgment

On first use, the agent surfaces the full security policy and asks you to type `I accept` before it performs any actions. This one-time gate ensures you understand the permissions and risks involved, including shell and filesystem access, potential cloud costs, and credential exposure. Once accepted, the gate doesn't appear again for that project. For the full security model, see [Anyscale agent skills security](/agent-skills/security.md).

---

Previous: [Overview](/agent-skills.md) | Next: [Security](/agent-skills/security.md)