Quick Start Guide

Get your first AI task running on runpiper in under 60 seconds.

Prerequisites

Before you begin, make sure you have:

  • A Unix-like operating system (Linux or macOS)
  • An internet connection
  • Basic familiarity with the command line

Windows users: runpiper works great with WSL2 (Windows Subsystem for Linux).

Installation

Install the runpiper CLI with a single command:

curl -fsSL https://codeberg.org/runpiper/runpiper/raw/branch/main/scripts/install.sh | sh

Or download binaries from Releases.

Verify the installation:

rp --version

Login to runpiper Cloud

rp auth login

This will open your browser to authenticate you with your runpiper account.

Create Your First Task

Let’s create a simple task:

# Create a new task
rp task init my-task

This creates a Taskfile template (my-task.toml) that you can customize.

Configure Your Task

Open my-task.toml to see the configuration:

[task]
name = "my-task"

[model]
provider = "anthropic"
name = "claude-3-haiku"

[prompts]
system = "You are a helpful assistant"
user = "{{input.message}}"

Edit the prompts to define what your task does.

Push to runpiper

Push your task to create a preview version:

rp task push my-task.toml

Test Your Task

Test your task with sample input:

rp task test my-task --input '{"message": "Hello, world!"}'

Deploy to Production

Deploy your task to production:

rp task deploy my-task

Your task is now live and ready to handle requests.

List Your Tasks

See all your tasks:

rp task list

Next Steps

Now that you have your first task running, here’s what to explore next:

  • Tasks: Learn about tasks and how to configure them
  • Capabilities: Connect to external services like databases and APIs
  • Agents: Build autonomous agents that use tasks and capabilities
  • CLI Reference: Complete CLI command documentation

Getting Help

Need help? We’re here for you: