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

# Builder Tabs

> A map of every tab in the agent builder — what it's for, what to fill in, and when to use it.

## What is this?

When you **Create** or **Edit** an agent, the builder opens with one tab per area of the agent — General, Behavior, Instructions, Knowledge, Capabilities, and more. Think of each tab as one part of onboarding a new employee: their name badge, their training manual, their reference binder, their system access.

This section explains every tab in plain language: what it's for, what to type or choose, and how it connects to the others.

<Tip>
  New to AI agents? Start with [Create Your First AI Agent](/workspace/ai-agents/create-your-first-agent) — a single walkthrough that builds one agent start to finish. Come back to these pages afterward for the full detail on any one tab.
</Tip>

## What can I do here?

| Tab                       | What it's for                                       | Page                                                                     |
| ------------------------- | --------------------------------------------------- | ------------------------------------------------------------------------ |
| **General**               | Name the agent and describe its job                 | [General](/workspace/ai-agents/tabs/general)                             |
| **Behavior**              | Set its tone, priority, and status                  | [Behavior](/workspace/ai-agents/tabs/behavior)                           |
| **Instructions**          | Write how it should think and act                   | [Instructions](/workspace/ai-agents/tabs/instructions)                   |
| **Knowledge**             | Give it facts to look things up in                  | [Knowledge](/workspace/ai-agents/tabs/knowledge)                         |
| **Capabilities**          | Turn on real actions it can take                    | [Capabilities](/workspace/ai-agents/tabs/capabilities)                   |
| **External Capabilities** | Connect your own APIs or MCP tools                  | [External Capabilities](/workspace/ai-agents/tabs/external-capabilities) |
| **Skills**                | Add a specific procedure for one situation          | [Skills](/workspace/ai-agents/tabs/skills)                               |
| **Workflows**             | View scripted, multi-step flows bound to this agent | [Workflows](/workspace/ai-agents/tabs/workflows)                         |
| **Channels**              | Choose where it runs and how it talks               | [Channels](/workspace/ai-agents/tabs/channels)                           |
| **Guardrails**            | Turn on safety checks                               | [Guardrails](/workspace/ai-agents/tabs/guardrails)                       |
| **Test**                  | Try it before customers do                          | [Test](/workspace/ai-agents/tabs/test)                                   |

## How to use it

### Save before you go further

**Capabilities**, **External Capabilities**, **Skills**, and **Workflows** only unlock after the agent has been created once — save on the General tab first.

### The pattern that repeats across tabs

One idea shows up again and again in this builder, so it's worth naming once: **almost every "when should this be used" decision is written on the thing itself, not in the agent's Instructions.**

* An automation's own **Description** decides when [Trigger Automation](/workspace/ai-agents/tabs/capabilities#trigger-automation) picks it — see [Capabilities](/workspace/ai-agents/tabs/capabilities).
* A skill's own **Description** decides when it activates — see [Skills](/workspace/ai-agents/tabs/skills).
* An external capability's own **"When should the AI use this tool?"** field decides when it's called — see [External Capabilities](/workspace/ai-agents/tabs/external-capabilities).
* Even a whole **specialist agent's** own Description decides when the platform routes a conversation to it — see [General](/workspace/ai-agents/tabs/general).

<Tip>
  Whenever you're setting up something the agent might use, ask: "did I write a clear, one-sentence description of exactly when this applies?" That sentence — wherever it lives — is doing more work than anything in the main Instructions box.
</Tip>

## Related docs

* [AI Agents Overview](/workspace/ai-agents/overview)
* [Create Your First AI Agent](/workspace/ai-agents/create-your-first-agent)
* [Automation Overview](/workspace/automation/overview)