AutoGPT: Platform for building autonomous AI agents
Block-based visual editor for autonomous AI agents.
Learn more about AutoGPT
AutoGPT is a platform for building and deploying autonomous AI agents through a visual programming interface backed by an execution server. The frontend implements a block-based editor where users compose agent behaviors by connecting discrete action nodes into workflows, while the backend server handles agent instantiation, lifecycle management, and execution of the defined behavior graphs. The platform provides abstraction layers for integrating multiple language model providers including OpenAI's GPT-4 and open-source alternatives like Llama, enabling model-agnostic agent development. Deployment options include containerized self-hosting via Docker or cloud-based hosting, with both supporting continuous operation modes where agents can be triggered by external events or run on schedules. The system includes a marketplace component for distributing pre-configured agent templates, facilitating reuse of common automation patterns across different deployment environments.
Visual Block-Based Builder
Agents are constructed by connecting discrete action blocks in a visual editor, eliminating code-based configuration. Each block performs a single operation, enabling rapid prototyping and modification without writing integration logic.
Flexible Deployment Options
Self-host via Docker with automated setup scripts or use cloud-hosted infrastructure. Supports the same agent definitions across environments, allowing migration between deployment models without reconfiguration.
Continuous Agent Execution
Deployed agents run persistently and respond to external triggers rather than terminating after single requests. Enables workflow automation and event-driven architectures without manual restart or polling mechanisms.
Major platform update with persistent file storage, speech-to-text via Whisper, and enhanced agent generation capabilities.
- –Add User Workspace for persistent CoPilot file storage
- –Copilot speech to text via Whisper model
- –Add library agent fetching with two-phase search for sub-agent support
- –Cover webhook ingress URL route
- –Add Redis-based SSE reconnection for long-running CoPilot operations
Feature-rich update with Claude Code execution, external Agent Generator, analytics integration, and improved chat system.
- –Add ClaudeCodeBlock for executing tasks via Claude Code in E2B sandbox
- –Add external Agent Generator service integration
- –Add PostHog analytics and OpenRouter tracing to chat system
- –Implement clarification questions UI for agent generation
- –Persist long-running tool results to survive SSE disconnects
Bug fixes for Sentry alert spam reduction and chat authentication flow improvements.
- –fix(platform): reduce Sentry alert spam for expected errors
- –fix(platform): evaluate chat flag after auth for correct redirect
Related Repositories
Discover similar tools and frameworks used by developers
Segment Anything
Transformer-based promptable segmentation with zero-shot generalization.
ComfyUI
Visual graph-based diffusion model workflow builder.
Ollama
Go-based CLI for local LLM inference and management.
CUTLASS
CUDA C++ templates and Python DSLs for high-performance matrix multiplication on GPUs.
Model Context Protocol Servers
Reference implementations for LLM tool and data integration.