> ## Documentation Index
> Fetch the complete documentation index at: https://learn.nexudus.com/llms.txt
> Use this file to discover all available pages before exploring further.

# What are Agent Skills?

> How Agent Skills let AI assistants like GitHub Copilot and Claude manage your Nexudus coworking spaces through natural language.

# Agent Skills

**Agent Skills** teach AI assistants how to use the Nexudus CLI. Once installed, your AI assistant (GitHub Copilot or Claude) can manage your Nexudus coworking spaces through natural language instructions — no need to remember command syntax.

## How it works

The Agent Skill is a structured knowledge file (`SKILL.md`) that gets installed into your AI assistant's context. It contains:

* **Command reference** — every CLI command, its options, and expected output.
* **Decision trees** — step-by-step workflows for common tasks (finding a business, creating a product, updating an entity).
* **Output parsing rules** — how to interpret the JSON envelope returned by `--agent` mode.
* **Error handling** — how to diagnose and recover from common errors.
* **Entity index** — a catalogue of all 40+ entity types the CLI supports.

When you ask your AI assistant something like *"List all products in my London office"*, the skill teaches the assistant to:

1. Run `nexudus businesses list --query "London" --agent` to find the business ID.
2. Run `nexudus products list --business <id> --agent` to list products for that location.
3. Parse the JSON envelope and present the results.

## Supported AI assistants

| Assistant              | Support                                                       |
| ---------------------- | ------------------------------------------------------------- |
| **GitHub Copilot**     | Fully supported — install the skill globally or per-workspace |
| **Claude** (Anthropic) | Fully supported via the same skill file                       |

## What the AI assistant can do

Once the skill is installed, your AI assistant can:

* **Query data** — list, search, and filter any entity type (businesses, products, resources, bookings, coworkers, invoices, etc.).
* **Create entities** — set up new products, resources, bookings, and more using natural language.
* **Update entities** — modify properties, assign tariffs, change pricing, upload images.
* **Delete entities** — remove entities with proper confirmation handling.
* **Run entity commands** — execute special operations like archiving or activating.
* **Diagnose issues** — run `nexudus doctor` and interpret the results.

## Example conversations

**You:** *"Show me all the meeting rooms in my space"*

The assistant runs:

```bash theme={null}
nexudus resources list --agent
```

and presents the results.

***

**You:** *"Create a day pass product for £25"*

The assistant runs `nexudus whoami --agent` to get your default business and currency, then:

```bash theme={null}
nexudus products create --name "Day Pass" --price 25.00 --business <id> --agent
```

***

**You:** *"Delete product 12345678"*

The assistant confirms the entity exists, then:

```bash theme={null}
nexudus products delete 12345678 --yes --agent
```

## With great power comes great responsibility

<Warning>
  Agent Skills give your AI assistant the ability to **create, update, and delete real data** in your Nexudus account. Before using them, be aware of the following risks:

  * **Destructive actions are real** — deletions and updates are applied immediately to your live environment. There is no undo button.
  * **AI can misinterpret intent** — a vague instruction like *"clean up old bookings"* could lead to unintended bulk deletions. Be specific in your requests.
  * **Always review before confirming** — read the commands and parameters the assistant proposes before letting them run, especially for write operations.
  * **Start in a test environment** — if your plan supports it, try commands against a sandbox or test business first.
  * **Limit scope when possible** — only grant the assistant access to the workspaces and businesses it needs.
</Warning>

## Important: the `--agent` flag

The skill instructs the AI assistant to always append `--agent` to every CLI call. This flag returns a structured JSON envelope that the assistant can parse reliably, instead of human-formatted table output.

## Next steps

<Card title="Install Agent Skills" icon="download" href="/cli/agent-skills-installation">
  Step-by-step installation guide for GitHub Copilot and Claude.
</Card>