Overview

Relevant source files

• .dockerignore
• .github/ISSUE_TEMPLATE/1.bug.yml
• .github/ISSUE_TEMPLATE/2.feature.yml
• .github/PULL_REQUEST_TEMPLATE.md
• .github/workflows/platform-backend-ci.yml
• .github/workflows/platform-frontend-ci.yml
• .github/workflows/platform-fullstack-ci.yml
• .gitignore
• CODE_OF_CONDUCT.md
• CONTRIBUTING.md
• LICENSE
• README.md
• autogpt_platform/README.md
• autogpt_platform/backend/.env.default
• autogpt_platform/backend/Dockerfile
• autogpt_platform/backend/README.advanced.md
• autogpt_platform/backend/README.md
• autogpt_platform/backend/backend/copilot/tools/agent_browser.py
• autogpt_platform/backend/backend/copilot/tools/agent_browser_integration_test.py
• autogpt_platform/backend/backend/copilot/tools/conftest.py
• autogpt_platform/backend/backend/util/feature_flag_test.py
• autogpt_platform/backend/test/__init__.py
• autogpt_platform/db/docker/docker-compose.yml
• autogpt_platform/db/docker/reset.sh
• autogpt_platform/docker-compose.platform.yml
• autogpt_platform/docker-compose.yml
• autogpt_platform/frontend/Dockerfile
• autogpt_platform/frontend/src/app/page.tsx
• autogpt_platform/frontend/src/services/feature-flags/__tests__/helpers.test.ts
• autogpt_platform/frontend/src/services/feature-flags/feature-flag-provider.tsx
• autogpt_platform/frontend/src/services/feature-flags/helpers.ts
• docs/content/code-of-conduct.md
• docs/content/contributing.md

This document introduces the AutoGPT repository, its dual-license structure, and the two major product lines it contains: the modern AutoGPT Platform and Classic AutoGPT. The AutoGPT Platform is a full-stack application for building, deploying, and managing AI agent workflows through a visual interface. Classic AutoGPT represents the original standalone agent approach with development tools like Forge and the agbenchmark testing framework.

For detailed information on specific subsystems, see:

• Repository Structure
• Core Components
• Key Terminology
• AutoGPT Platform
• Classic AutoGPT
• Getting Started
• Development

Repository Structure

The AutoGPT repository is organized into two independently-licensed product lines that share minimal code but coexist in the same repository for historical and community reasons.

Directory Layout

Sources: README.md127-134 .gitignore3-17

License Structure

The repository employs a dual-license model:

Component	License	Path	Description
AutoGPT Platform	Polyform Shield	autogpt_platform/	Modern agent builder platform with visual workflow editor
Classic AutoGPT	MIT	classic/	Original standalone agent, Forge toolkit, and agbenchmark
Shared Infrastructure	MIT	Root level files	CI/CD workflows, documentation, CLI tools

The Polyform Shield License restricts commercial use (providing a competing product) without a separate commercial license, while the MIT License permits unrestricted use.

Sources: LICENSE1-19 README.md127-134 CONTRIBUTING.md10-11

AutoGPT Platform Architecture

The AutoGPT Platform is a complete rewrite of the agent concept as a visual workflow builder. Users create "agent graphs" by connecting "blocks" (nodes) that perform specific actions. It also includes a Copilot system for AI-assisted agent creation and interaction.

Platform Overview Diagram

Sources: autogpt_platform/docker-compose.platform.yml9-35 README.md98-107

Key Components

Frontend (autogpt_platform/frontend/)

A Next.js application featuring a low-code Agent Builder and a Copilot chat interface. It uses React Flow for graph visualization and Supabase for authentication.

• For details, see Frontend.

Backend (autogpt_platform/backend/)

A microservice-based Python application. Key services defined in the platform configuration include rest_server, executor, websocket_server, and scheduler_server.

• For details, see Backend.

Copilot System

An AI-powered interface that allows users to build and run agents via natural language. It supports two execution paths: a Claude Agent SDK path and a baseline OpenAI-compatible path.

• For details, see Copilot System.

Shared Libraries (autogpt_platform/autogpt_libs/)

Internal Python utilities for authentication, logging, and feature flags shared across backend services.

• For details, see Shared Libraries (autogpt-libs).

Sources: autogpt_platform/docker-compose.yml70-111 README.md80-96

Classic AutoGPT Components

Classic AutoGPT represents the original MIT-licensed components focused on traditional autonomous agent development and benchmarking.

Classic Architecture Diagram

Sources: LICENSE13-17 README.md132-134

• Forge: A toolkit for building agents following the Agent Protocol standard. See Forge.
• agbenchmark: A standardized testing framework for autonomous agents. See Benchmark (agbenchmark).

External Service Dependencies

The AutoGPT Platform integrates with several infrastructure services:

Service	Purpose	Source Reference
Supabase	Auth & User Management	autogpt_platform/docker-compose.yml139-152
PostgreSQL	Primary Data Store	autogpt_platform/docker-compose.platform.yml33-34
Redis	Distributed Locking (3-shard cluster)	autogpt_platform/docker-compose.platform.yml100-137
RabbitMQ	Message Broker	autogpt_platform/docker-compose.platform.yml167-175
FalkorDB	Graph Memory Store	autogpt_platform/docker-compose.platform.yml138-166
ClamAV	Security / Virus Scanning	autogpt_platform/docker-compose.yml112-132

System Requirements

• CPU: 4+ cores recommended
• RAM: 8GB Minimum (16GB recommended)
• Software: Docker Engine (20.10.0+), Docker Compose (2.0.0+), Node.js (16.x+), Python (3.11+).

Sources: README.md27-52 autogpt_platform/backend/Dockerfile24-34