Quickstart

Learn Endstack's essential navigation, prepare Chrome for authenticated work, start a channel, and configure your workspace.

Endstack brings your desktop, channels, agent, files, and settings into one workspace. This tour covers the navigation and setup you need before you start working.

1. Learn the basic navigation

The endOS desktop is your workspace surface inside Endstack. You can reach the same desktop and open apps from the web, desktop, and mobile clients.

Toggle the channel pane using the Endstack button in the dock.

On desktop

The Endstack logo at the left of the dock controls the channel pane. Click it to open or close Channels. Your desktop remains available beside the pane, so you can move between a conversation and your apps without leaving the workspace.

On mobile

  • If an app is open, tap the Endstack logo to return to the desktop. From the desktop, tap it to open or close Channels.
  • Press and hold the logo to open the app switcher, an iOS-style carousel of your open apps.
  • Swipe right from the left edge to bring in Channels or return from a channel to the channel list. Swipe left from the right edge to dismiss Channels and return to the desktop.

See Desktop overview for the rest of the desktop controls.

2. Sign in to Chrome

Chrome in Endstack is a full, persistent browser, not a disposable session. Your browser profile, cookies, and signed-in site sessions remain available when you close Chrome or reconnect to your desktop.

Open Chrome and sign in to your Chrome profile, the individual websites you use, or both. The Endstack agent can then use Chrome to work with authenticated sites available in that browser session, such as dashboards, admin tools, and web apps.

Learn more in Search and browser use and Native apps.

3. Set up your identity

Open Settings → Profile and add the display name and profile picture you want teammates to see. Your profile appears in channels, messages, mentions, and member lists.

If you manage the workspace, open Settings → Members to invite teammates as Members or Admins.

4. Start a channel

Open Channels from the Dock. Channels are shared conversations where workspace members and the Endstack agent can work together.

In a channel, try a request with a concrete outcome:

text
Research the pricing pages of our top five competitors.
Save a comparison to ~/research/pricing.md and summarize the key differences here.

Agent status, approval requests, and the final response appear in the channel where the run started. If a tool needs confirmation, review its in-channel card and choose Approve or Deny.

Learn more in Channels overview and Autonomy and approvals.

5. Work with your files

Each member has a private 50 GB Linux home directory. The Files app, Terminal, EndCode, Writer, and the agent all work with the same files.

Files gives you one view of your private files, Team Drive, and items shared with you.

If you like using the terminal, you can create even a folder from Terminal and open the result immediately in Files:

bash
mkdir -p ~/researchprintf "# Pricing notes\n" > ~/research/pricing.md

Otherwise, you can right click on the desktop to create files or folders.

Use Team Drive for workspace-wide material, or share individual files and folders with specific teammates. Files shared through a channel also appear in Shared with me for channel members.

6. Configure the agent

Use Settings to shape how the agent works:

Settings sectionWhat to configure
AgentTool autonomy and the AI review policy
ModelsFast and Pro models, plus workspace provider credentials
SkillsPrivate and workspace-shared instruction packs
MCPLocal or remote MCP servers
IntegrationsThird-party apps, Email, Messaging, and Slack

Start with Agent overview, then connect only the capabilities your work needs.

7. Automate a recurring workflow

Create a task channel when work should run once at a later time or on a recurring schedule. A task can research, write files, post to an endOS channel, send email, message Slack, or act through a connected app.

Keep Whitelist task tools off while designing a task. Turn it on only after you trust the instructions and want scheduled runs to execute without per-tool approvals.

See Tasks for examples and delivery options.

Next steps