> For the complete documentation index, see [llms.txt](https://docs.dapta.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.dapta.ai/dapta-mcp/key-concepts.md).

# Key Concepts

Before you connect, it helps to understand four ideas. They explain how the MCP authenticates, how it decides what to do, and how it keeps your data safe.

***

## 1. One API key per workspace

A single active key authenticates your AI client as **one specific workspace**. The MCP automatically resolves which workspace you are in from the key, so you never pass a workspace ID by hand.

* **One active key per workspace.** Revoke the existing key before minting a new one.
* **Shown once.** Keys are displayed only at the moment you create them. Save the key immediately.
* **Owner-only.** Only workspace owners can mint or revoke keys.

{% hint style="warning" %}
Treat your key like a password. If it is ever exposed, revoke it and mint a new one.
{% endhint %}

***

## 2. Tools are the actions

**Tools** are the raw, concrete actions Claude can take: read a list of agents, fetch call statistics, create a flow. Each tool does one thing. The MCP ships **128 tools** grouped into 13 domains. You can browse them all in the [Tools Reference](/dapta-mcp/tools-reference.md), but you will rarely need to: the skills pick the right tools for you.

***

## 3. Skills are the playbooks

**Skills** are guided workflows that tell Claude *how* and *when* to use tools for a real task. You do not invoke a skill by name. You describe what you want in plain language, and the matching skill activates automatically, then orchestrates the right tools in the right order.

Tools and skills ship together: baked into the plugin for Claude Desktop and Cowork, or as a downloadable bundle for Claude Code. The [Skills Catalog](/dapta-mcp/skills.md) lists all 33 with example phrases to try.

***

## 4. Read versus write: the safety model

This is the most important concept for trusting the MCP with your account.

* **Read tools** return data directly. Asking "how many calls did I get this week?" runs a read and shows you the answer.
* **Write tools use a two-step preview, then commit pattern.** When Claude is about to change something, it first shows you exactly what it will change and returns a preview. **Nothing is written until you confirm and a matching commit step runs.** This prevents accidental or surprising changes.

{% hint style="info" %}
The one exception is creating a brand-new voice agent, which is created directly once you have supplied the details. Every other write (updating an agent, editing a prompt, building or running a flow, opening a support ticket) goes through preview, then commit.
{% endhint %}

***

## Next step

Ready to connect? The setup section walks you through minting your key, connecting your client, and (for Claude Code) installing the skills.

{% content-ref url="/pages/MAXWVsEp5drMKkYRSLZw" %}
[Setup & Connection](/dapta-mcp/setup.md)
{% endcontent-ref %}


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.dapta.ai/dapta-mcp/key-concepts.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.