Usage Guide

Complete command reference and operational modes for Labs CLI.

1. Operational Modes

Labs operates in several distinct modes depending on your workflow requirements.

Interactive TUI (Default)

The primary interface. Launches a full-screen Terminal User Interface with chat, history, and real-time thinking visualization.

labs

Desktop Mode

A variant of the TUI optimized for desktop environments, often enabling deeper OS integrations or specific display rendering preferences.

labs desktop

Telegram Bot Bridge

Connects Labs to a Telegram bot, allowing you to interact with your cognitive agent remotely via chat. Ideal for mobile access or asynchronous workflows.

Configuration
labs telegram
Run Bridge
labs telegram run

2. Knowledge & Memory

Commands for managing the agent's long-term semantic memory (HoloMem).

study <file>

Ingests a local file (Markdown, Code, PDF, etc.) into the vector database.

labs study ./README.md

recall <query>

Performs a semantic search against the vector memory to retrieve relevant context.

labs recall 'architecture diagrams'

dive <docId>

Inspects the full content and metadata of a specific document stored in memory by its ID.

labs dive doc-12345

reset brain

Warning: Wipes all long-term memory and vector embeddings. Requires the --force flag to execute.

labs reset brain --force

3. Research & Utilities

Tools for web access and low-level system testing.

search <query>

Performs a Google search and returns summarized results directly in the terminal.

labs search 'latest nextjs features'

browse <url>

Visits a webpage, renders it headless, and extracts the main content for reading.

labs browse https://example.com

sdk <tool> [args]

Direct access to the internal toolchain. Useful for testing specific capabilities (file ops, git, system checks) in isolation.

labs sdk --list
labs sdk system_info