Getting Started

Local AI orchestration for extracting information and creating new content

Best Performance

For best performance, use either an Apple Silicon Mac (M1 or newer) or an NVIDIA GPU with at least 4 GB of VRAM and CUDA 12.4.

Larger GPUs allow orchestration to select larger models for better task quality and performance, instead of defaulting to smaller models such as a 1B parameter model.

Installation

macOS / Linux:

curl -fsSL https://smartloop.ai/install.sh | bash

Windows (PowerShell):

irm https://smartloop.ai/install.ps1 | iex

Uninstall

macOS / Linux:

curl -fsSL https://smartloop.ai/uninstall | bash

Windows (PowerShell):

irm https://smartloop.ai/uninstall.ps1 | iex

Usage

To verify the installation, please type ```slp --help``:

usage: slp [-h] [--version] [--resume CONVERSATION_ID] [-C PATH]
               {run,status,agent,token,skill,mcp,project} ...

AI Agent harness for extracting information and creating new content

positional arguments:
  {run,status,agent,token,skill,mcp,project}
                        Available commands
    run                 Start interactive chat session
    status              Show project status
    agent               Agent management commands (start/stop/status/restart)
    token               Manage your developer token
    skill               Manage skills for the active project
    mcp                 Manage MCP servers for the active project
    project             List, current, or switch the current project

options:
  -h, --help            show this help message and exit
  --version             show program's version number and exit
  --resume CONVERSATION_ID
                        Resume a previous conversation by session ID
  -C PATH, --path PATH  Working directory to use (defaults to ~/smartloop or
                        ~/smartloop/<project>) 

Use it to extract information from your personal documents and do tasks for you, like writing guidelines, creating recipes, etc. Build up skills for more specialized tasks like creating events, tickets, connecting external systems, or summarizing a blog post.

Your projects live inside the ~/smartloop/<slug> directory, and this is where skills and documents are copied.

Use the CLI commands to complement your Studio app experience. For example, if you have uvx installed on your machine, you can run the following command to register a local MCP server:

slp mcp add uvx yahoo-finance-mcp

Requirements

Requirement	Description
OS	macOS (Apple Silicon), Linux (x86_64), or Windows (x86_64)
CUDA	12.4+ (optional - NVIDIA GPU acceleration on Linux/Windows, defaults to CPU)