> ## Documentation Index > Fetch the complete documentation index at: https://docs.splox.io/llms.txt > Use this file to discover all available pages before exploring further. # The workspace > The sidebar, the machine switcher, the composer, and where your chats live Everything you do in Splox happens on one screen: a sidebar of places on the left, a conversation in the middle, and a composer at the bottom. This page names the parts so the rest of this section can point at them. The Splox workspace with an empty chat open ## The sidebar **New Chat** opens an empty conversation. **Machines**, **Tools**, **Skills** and **Connections** are the four screens where you set up what the agent has to work with: | Where | What it is | | ------------------------------- | ---------------------------------------------------------------------------------------- | | [Machines](/app/machines) | The computers your agents run on, the harnesses they run, and the starters you can copy. | | [Tools](/app/tools) | The system tool packages and the MCP servers you have connected. | | [Skills](/app/skills) | The skills catalog, the ones the platform ships, and your own. | | [Connections](/app/connections) | Model providers you bring yourself, and environment secrets. | Under those, **PROJECTS** lists your projects and **RECENT** lists your chats, newest first. A chat sits in Recent until you move it into a project, and then it sits under that project instead. See [Projects](/app/projects). The magnifier at the top of the sidebar searches chats — by name and by message content, so a phrase you remember from a conversation finds it. At the bottom is your account: name, plan, and a menu with **Upgrade Plan**, **Usage**, **Settings** and **Log out**. The download icon beside it goes to the [desktop app](/app/desktop). ## The machine switcher The control at the top left of a chat names the **machine** this conversation runs on, and under it the harness that machine runs and the ref it runs at. The machine picker open in a chat This is not an agent picker. Switching machines switches computers: a different disk, a different checkout of a possibly different harness. A chat stays on the machine it was started on, so switch before you send the first message rather than in the middle of a conversation. [Machine](/concepts/machine) explains what that means for your files. The picker also says whether the machine is running, and ends with a **Your machines** shortcut to the [Machines](/app/machines) screen. ## The composer The composer with the run configuration popover open Three controls: * **+** attaches a file. It is uploaded and the agent is given a URL it can fetch onto the machine. * **Harness decides · Default** is the run configuration: which model answers, and how much thinking effort it spends. Left alone, both come from the code in your harness. * The arrow sends. While a run is going it becomes a square, which stops it. All three are covered in [A conversation](/app/chat). ## The inspector The button at the top right of a chat opens a panel showing the machine this chat is on, how much of the context window the conversation has used, a way into the sandbox over SSH or VS Code, and any sub-agents running right now. The chat inspector panel open beside a finished run ## Where to go next Sending, attachments, reading tool calls, stopping, forking, sharing. Start, stop, add a second one, delete one. What is listed, what `system:` means, how an agent gets one. The catalog, the shipping skills, forking one.