Key Features
What’s common to tools like OpenClaw
- Plugins — Extensible plugin system for tools, channels, hooks, and skills
- Channels — Interact via TUI (built-in), Telegram (plugin), HTTP API, or custom channels
- Skills — Read-only knowledge packages for agent context
- Policy Enforcement — Fine-grained control over what agents can do
- Audit Logging — Every tool invocation is logged for accountability
What’s different
- Tools become CLI executables — Agents write scripts that call tools directly, eliminating round-trips through the LLM
- Docker Sandboxing — always — All code execution happens in isolated containers with no escape hatch. Communication to the gateway via sockets.
- Nothing out-of-the-box — You configure agents the way YOU want
Why?
Why build a new agent system if Openclaw, Picoclaw and others already exist? What’s so different with Beige? Find out on Why Beige.How It Works
Beige uses a two-host model: The Gateway is the orchestrator. It loads plugins, manages Docker containers, routes tool calls, enforces policies (deny by default), logs every action, and serves channels. The Sandbox is where each agent runs — an isolated Docker container with a writable/workspace, plugin tools on $PATH, read-only skill mounts, and no access to host secrets or environment variables.
Next Steps
Installation
Install Beige, configure your first agent, and run it
Why Beige
The motivation, inspiration, and use cases behind Beige
The Gateway
Deep dive into architecture and the security model
Config Reference
Complete config.json5 reference — all fields and validation rules