Slaves
Agents
Autonomous task execution
Slaves can operate as autonomous agents — executing multi-step tasks without constant supervision.
What agents can do
- Execute shell commands
- Browse the web (Playwright)
- Read and write files
- Call APIs
- Chain multiple tools together
- Handle long-running workflows
Enabling agent mode
Agent mode is enabled by default. When a user sends a complex task, the slave automatically enters agent mode and works through it step by step.
Tools
Agents have access to a toolkit:
| Tool | Description |
|---|---|
| Shell | Execute commands |
| Browser | Navigate and interact with web pages |
| File | Read and write files |
| Search | Web search |
| Memory | Store and recall information |
Safety
Agent actions can be configured with permission levels:
- Auto — execute without confirmation
- Confirm — ask the user before executing
- Deny — never execute
Configure in your slave's config file under the agent.permissions section.