Desktop app

Run agent sessions in a workspace

Reviewed by

Updated Sep 10, 2026 4 min

A workspace is a folder you work in, with three things attached: a file tree, its git history, and a launcher for the agent CLIs you already use. Add an existing project folder to launch agents and review their changes in the same app.

Add a workspace

Point the app at a folder. It reads the git state and the file tree, and the folder appears in the sidebar under the machine it lives on. Nothing is copied and nothing is rewritten: it is your working copy, opened.

Start a session

Choose an installed agent from the launcher and review its model, flags, and permissions. Each profile runs that agent's own CLI in the project folder. The launcher also has a plain shell and a file browser. See supported tools for launch and collection coverage.

The agent uses its own subscription or API credentials. A launch profile does not replace its permission settings or prevent it from running Git commands.

The Configure Grok Build dialog over the launcher: the binary and config paths it found, a model field set to grok-4.6, effort buttons from low to xhigh, a compaction slider, and permission modes from default to always-approve — Each profile keeps its own model, effort and permission settings, saved to that tool's own config.

A session is a tab, not a tool, so two Claude Code sessions on two branches are two tabs. Several run side by side in one workspace, and the sidebar shows what each of them is doing without you having to click into it.

Sessions belong to the host process. Closing a window can leave them running while the host remains active. Quitting the app, stopping the service, sleeping, and shutting down the machine are different actions.

Enable Always-on host if work should continue after the desktop app quits. Keep the machine awake and reachable for remote access. See host setup.

Read the usage

Workspace and usage views connect recorded activity to its project. Amounts may be estimated API list-rate value, including subscription-covered work. They are not necessarily additional charges. Coverage depends on what the agent records and when it was collected.

Review before you commit

The History panel shows uncommitted changes with a real diff per file, a change count, and a commit box. Use the review controls to choose what to stage and commit. Agents and command steps can also change Git state according to their permissions and instructions. The review UI does not block those actions.

The History panel beside a session: a list of commits on the right, and on the left one commit opened up with its message, its file count, and a syntax-highlighted diff with added lines marked — A commit opened from the History list, with its message and the diff that went in.

Files open in a file editor beside the sessions: syntax highlighting, line numbers, find, bracket pairing, and changed-line marks in the gutter read from the diff the app already parsed. Saving is explicit. For language-server features or completion, open the file in your preferred editor.

Very large files open unhighlighted and say so, rather than opening slowly.

A shell script open in the tokenstat editor, with line numbers and syntax highlighting, the workspace file tree on the right, and the session tabs still along the top — A file open beside the sessions, with the tree on one side and the tabs still where you left them.

The desktop app covers the other surfaces in the window.
Set up the tokenstat desktop app if you have not installed it yet.
Let an agent look up its own spend does the same job from the other direction: the agent asks tokenstat.

Let an agent find this workspace

Through tokenstat MCP, an agent can call workspace_list to discover folder IDs and paths, then workspace_context for tasks and notes in one folder or all workspaces. Use the returned workspaceId when creating project tasks or memos. See the setup guide and copyable agent prompt.