OpenClaw Glossary

An autonomous AI program capable of performing actions — managing files, calling APIs, controlling applications. OpenClaw is a platform for creating a...

A unique token for accessing a language model's API (Claude, GPT, DeepSeek). Stored in the OpenClaw configuration file and never shared with third par...

An OpenClaw feature for controlling a web browser — scraping pages, filling forms, taking screenshots, navigating websites via AI commands. Browser Re...

A browser extension that connects your Chrome browser to OpenClaw. It allows the AI agent to read page content, fill forms, click buttons, and automat...

The official marketplace for OpenClaw skills. You can discover, install, and publish skills here. Similar to npm for Node.js packages, ClawHub provide...

An earlier name for the OpenClaw project during its development phase. Clawdbot evolved from Moltbot, incorporating more advanced agent capabilities a...

The OpenClaw command-line tool for installation, configuration, skill management, and diagnostics. Key commands: install, config, pair (connect messen...

The main configuration file for an OpenClaw instance in JSON format. Contains model settings, Gateway port, API keys, installed skills list, and works...

A scheduled task system in OpenClaw for executing actions on a timetable — daily reports, website monitoring, automated checks. Configured in the Open...

A background process that keeps OpenClaw running 24/7 without user intervention. Managed via systemd on Linux or launchd on macOS. The daemon automati...

The web-based management interface for OpenClaw, also known as Mission Control. Provides real-time monitoring of agent activity, message logs, skill s...

A Chinese AI company offering high-quality language models at low prices. DeepSeek V3 and R1 are popular choices for OpenClaw due to their excellent p...

A containerization platform used to run OpenClaw in an isolated environment. Docker simplifies deployment and updates — one command (docker compose up...

The central process of OpenClaw that receives messages from messengers, routes them to the AI model, and returns responses. Runs on port 18789 by defa...

A periodic signal from the OpenClaw Gateway confirming the agent is alive and responding to requests. Used for monitoring and automatic restart on fai...

A popular budget hosting provider frequently used for OpenClaw deployments. Their KVM VPS plans start at $5/month and include enough resources for a b...

A large language model — the AI brain behind OpenClaw. Examples: Claude (Anthropic), GPT (OpenAI), DeepSeek, Llama (Meta). An LLM takes a text prompt ...

Apple's compact desktop computer, popular as a home server for OpenClaw. The M-series chips provide excellent performance for local AI models via Olla...

A protocol for connecting external tools and data sources to AI models. OpenClaw uses MCP servers for integrations with databases, file systems, APIs,...

OpenClaw's persistent knowledge system that stores important information across conversations. Memory allows the agent to remember user preferences, p...

The web dashboard for OpenClaw that provides a comprehensive overview of your agent's activity. Shows message history, active skills, token usage, sys...

A specific AI model used by OpenClaw to generate responses. Each model differs in quality, speed, and cost — Claude Opus 4 delivers the best quality, ...

The original project that eventually became OpenClaw. Created by Peter Steinberger, Moltbot started as a personal Telegram bot experiment and grew int...

A lightweight, minimal fork of OpenClaw designed for resource-constrained environments. NanoClaw strips away advanced features like multi-messenger su...

A JavaScript runtime required for OpenClaw to function. Minimum version: 22 LTS. Installed automatically by the OpenClaw setup script via nvm (Node Ve...

An authentication protocol used by some OpenClaw skills to connect to third-party services like Google, GitHub, and Slack. OAuth allows the agent to a...

A tool for running AI models locally without API keys. Supports Llama, Mistral, Qwen, Gemma, and dozens of other open models. Free to use but requires...

The initial setup wizard that guides you through choosing a model, entering an API key, and connecting your first messenger. Onboarding launches autom...

An open-source AI agent platform with 250K+ GitHub stars. Connects language models to messengers and tools, turning AI into a personal assistant. Self...

A multi-provider API gateway — one API key gives access to Claude, GPT, DeepSeek, and hundreds of other models. Single billing and unified request for...

The process of connecting a messenger to OpenClaw. For Telegram: provide the bot token from @BotFather. For WhatsApp: scan a QR code. Executed via the...

A text instruction sent to the language model. In OpenClaw, the prompt is formed automatically from the user's message, soul.md contents, current cont...

A company that provides a language model API — Anthropic (Claude), OpenAI (GPT), DeepSeek, Google (Gemini), Meta (Llama). In OpenClaw, the provider is...

An installable plugin that extends OpenClaw's capabilities. Examples: web-search, email, github, calendar. Installed via openclaw skill install <name>...

A configuration file that defines a custom skill's capabilities, tools, and instructions in Markdown format. SKILL.md is placed in the skill's directo...

A Markdown file that defines the personality, knowledge, restrictions, and behavior rules of your AI agent. It's the 'brain' and character of your Ope...

A secure remote access protocol for connecting to your server. Used for installing and managing OpenClaw on a VPS. SSH provides encrypted connections ...

Minimum hardware and software specs for running OpenClaw: 2 GB RAM, 1 vCPU, 20 GB SSD, Node.js 22+, Ubuntu 22.04+ or macOS 13+. For local models via O...

A unit of text processed by the language model. One token is approximately 4 characters in English. Payment for AI model usage is calculated based on ...

A text-based graphical interface for OpenClaw right in the terminal. Provides a chat-like experience with the agent from the command line without need...

A virtual server rented from a hosting provider (Hostinger, DigitalOcean, Hetzner). The recommended way to run OpenClaw 24/7 for stable operation. Bas...

An isolated environment in OpenClaw with its own config, memory, and skill set. Workspaces allow running multiple independent agents on a single serve...

A Windows feature that lets you run a Linux environment natively within Windows. Required for running OpenClaw on Windows, since the platform needs Li...

An ultra-minimal configuration of OpenClaw that runs with zero skills and zero external dependencies. ZeroClaw mode is useful for testing, debugging, ...