Getting Started

This guide walks you through creating your account, setting up a workspace, and running your first AI agent.

1. Create Your Account#

Visit the signup page and provide:

  • Workspace name — Your team or organization name
  • Workspace slug — URL-friendly identifier (3-63 characters, lowercase letters, numbers, and hyphens). This becomes part of your workspace URL and cannot be changed after creation.
  • Email address — You'll need to verify this
  • Password — Minimum 8 characters with at least one uppercase letter, lowercase letter, digit, and special character

After signing up, check your email for a verification link. You can resend the verification email from the login page if needed.

2. Configure an AI Provider#

Before running agents, you need to add credentials for at least one AI provider. RedStick AI supports providers like Anthropic, OpenAI, and others — you bring your own API key (BYOK), so you are never locked into a single vendor. Go to Providers from the sidebar and either:

  • Paste an API key directly (stored encrypted with AES-256-GCM encryption)
  • Connect via OAuth — Use the Anthropic OAuth PKCE flow to connect your Anthropic account without copying a key manually

The credential is encrypted at rest and never displayed in full after saving.

3. Create Your First Project#

From the dashboard, click New Project and configure:

  • Name — A descriptive project name
  • Description (optional) — What this project is about
  • Framework — Choose from 15 supported frameworks:
CategoryFrameworks
JavaScriptNext.js, Vite, React, Vue.js, Node.js, Express.js
PythonPython, Flask, FastAPI
GoGo, Fiber, Gin
RubyRuby, Rails
OtherNone (static HTML/CSS/JS)

The framework determines which container image, default commands, and agent instructions are used for previews and executions.

4. Run Your First Execution#

Open your project to enter the Command Center — the IDE-like workspace. Click Start Task to open the task dialog:

  1. Write a prompt — Describe what you want the worker to build or fix:

    "Create a landing page with a hero section, features grid, and footer"

  2. Plan mode — Defaults to always: the worker proposes a plan and waits for your approval before writing code. Change to never to skip the gate, or auto to let the worker decide. See Plan Mode.

  3. Advanced options (optional) — Override the default AI provider and model for this execution.

Click Start to launch the execution. The worker streams in real-time:

  • Real-time streaming of the agent's thinking and actions
  • Tool calls showing file creation, editing, and terminal commands
  • Diff views for code changes

After the worker starts, you can follow up by sending a new Run from the Runs tab.

5. Preview Your Work#

Once the agent finishes (or while it's running), click Start Preview in the right panel. This launches a container running your project with the appropriate framework-specific dev server.

The preview auto-refreshes as the agent makes changes. You can also share preview links with others.

6. Next Steps#