Files
AutoGPT/docs/platform/agent-blocks.md
Nicholas Tindle a318832414 feat(docs): update dev from gitbook changes (#11740)
<!-- Clearly explain the need for these changes: -->
gitbook branch has changes that need synced to dev
### Changes 🏗️
Pull changes from gitbook into dev
<!-- Concisely describe all of the changes made in this pull request:
-->

<!-- CURSOR_SUMMARY -->
---

> [!NOTE]
> Migrates documentation to GitBook and removes the old MkDocs setup.
> 
> - Removes MkDocs configuration and infra: `docs/mkdocs.yml`,
`docs/netlify.toml`, `docs/overrides/main.html`,
`docs/requirements.txt`, and JS assets (`_javascript/mathjax.js`,
`_javascript/tablesort.js`)
> - Updates `docs/content/contribute/index.md` to describe GitBook
workflow (gitbook branch, editing, previews, and `SUMMARY.md`)
> - Adds GitBook navigation file `docs/platform/SUMMARY.md` and a new
platform overview page `docs/platform/what-is-autogpt-platform.md`
> 
> <sup>Written by [Cursor
Bugbot](https://cursor.com/dashboard?tab=bugbot) for commit
e7e118b5a8. This will update automatically
on new commits. Configure
[here](https://cursor.com/dashboard?tab=bugbot).</sup>
<!-- /CURSOR_SUMMARY -->

<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->

## Summary by CodeRabbit

* **Documentation**
* Updated contribution guide for new documentation platform and workflow
  * Added new platform overview and navigation documentation

* **Chores**
  * Removed MkDocs configuration and related dependencies
  * Removed deprecated JavaScript integrations and deployment overrides

<sub>✏️ Tip: You can customize this high-level summary in your review
settings.</sub>

<!-- end of auto-generated comment: release notes by coderabbit.ai -->

---------

Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-09 19:22:05 +00:00

3.3 KiB

How to Create an AI Agent as a Block in AutoGPT

Overview

This guide explains how to create a reusable agent block that can be used as a component in other agents.

What Are Agent Blocks?

Agent blocks are pre-configured, reusable AI workflows that can be used as components within larger automation systems. Think of them as "smart building blocks" - each agent block is itself a complete workflow that can:

  • Accept specific inputs
  • Process data using AI and traditional automation
  • Produce defined outputs
  • Be easily reused in different contexts

The power of agent blocks lies in their modularity. Once you create an agent with a specific capability (like translating text or analyzing sentiment), you can reuse it as a single block in other workflows. This means you can:

  • Combine multiple agent blocks to create more complex automations
  • Reuse proven workflows without rebuilding them
  • Share agent blocks with other users
  • Create hierarchical systems where specialized agents work together

For example, a content creation workflow might combine several agent blocks:

  • A research agent block that gathers information
  • A writing agent block that creates the initial draft
  • An editing agent block that polishes the content
  • A formatting agent block that prepares the final output

Creating the Base Agent

Required Components

  1. Input Block
  2. AI Text Generator Block
  3. Output Block

Step-by-Step Setup

  1. Add and Configure Blocks
    • Add an Input Block
    • Add an AI Text Generator Block
    • Add an Output Block
  2. Connect Components
    • Connect Input's result to AI Text Generator's Prompt
    • Connect AI Text Generator's response to Output's value
  3. Name the Components
    • Name the Input Block: "question"
    • Name the Output Block: "answer"
  4. Save the Agent
    • Choose a descriptive name (e.g., "Weather Agent")
    • Click Save

Converting to a Block

  1. Access the Block Menu
    • Go to the Builder interface
    • Click the Blocks menu
    • Click the agent tag or search the name of your agent
  2. Using the Agent Block
    • Click on the agent block to add to your workflow
    • Save the new agent with a descriptive name (e.g., "Weather Agent")

Testing the Agent Block

  1. Run the Agent
    • Enter a test question (e.g., "How far is the Earth from the Moon?")
    • Click Run
  2. View Results
    • Option 1: Check "Agent Outputs" section*
    • Option 2: Click "View More" for detailed results

*Note: if there is no output block then the "Agent Outputs" button will show up blank. You can see the output under view more or at bottom of the block.

Advanced Usage

  • You can make more complex agents by combining multiple agent blocks
  • Chain different agents together for more sophisticated workflows

Note

This is a basic example that can be expanded upon to create more complex agent blocks with additional functionality.