Architecture Diagram

This interactive diagram breaks down the OpenClaw architecture into its core components: Inputs, the OpenClaw Core (Gateway & Orchestrator), the Brain (LLM), the Tools/Agents, and the Output.

OpenClaw System Architecture

A structural breakdown of inputs, core processing, reasoning, tools, and outputs.

Inputs & Triggers

The Users

Power Users and Developers

User Channels

WhatsApp, Telegram, Discord, iMessage

Cron Jobs & Heartbeat

Scheduled tasks, background monitoring, time-based triggers

The Brain

Cloud LLM

The reasoning engine. Does all the thinking, planning, and logic (e.g., Claude, GPT-4o, Ollama).

OpenClaw Core

Gateway

Integrates with channels, intercepts messages, and handles I/O.

Orchestrator

The framework. Manages complex tasks, multi-step workflows, and interacts with sub-agents.

Tools, Sub-Agents & Skills

Sub-Agents

Specialized workers (e.g., Coder, Researcher, Data Agent).

Skills & Toolbox

File System, Web Search, APIs, Python Execution.

Output

Actions

Email files, send messages to different users, execute code.

Generated Output

Text, Images, Voice, Code.

Read/Present Data

Summaries, spreadsheets, parsed documents.

Explore the diagram above to understand how data flows through the system and how complex tasks are managed.