TL;DR: Docker Desktop is not running, so the named pipe the CLI talks to does not exist. Start Docker Desktop from the Start menu and wait until the whale icon stops animating, then retry your command. The CLI on Windows talks to the engine through a Windows named pipe, and that pipe only exists while the Desktop backend is up.

## The error

```text
error during connect: Get "[url-encoded npipe address]/v1.47/containers/json": open //./pipe/dockerDesktopLinuxEngine: The system cannot find the file specified.
```

## Fix it

1. Start Docker Desktop from the Start menu.
   Expected: the whale icon appears in the system tray, animating while it boots.
2. Wait for the icon to stop animating (backend fully up, can take a minute on first boot).
3. In a fresh terminal, run:
   `docker ps`
   Expected: container list, no connect error.

## When this applies
- Windows 10/11 with Docker Desktop installed, commands fail with the npipe error
- Right after boot before Desktop has auto-started

## When this does NOT apply
- Linux (different socket, different fix)
- Desktop is running but commands still fail (check WSL2 backend health or switch container mode)
- `docker context ls` shows a non-desktop context selected

## Versions
Docker Desktop 4.x on Windows 10/11, both WSL2 and Hyper-V backends.

## Why it happens
The Windows CLI defaults to the `desktop-linux` context, which points at the `//./pipe/dockerDesktopLinuxEngine` named pipe. The pipe is created by the Desktop backend process; if that process is not running, every docker command fails at connect time.

## Edge cases
- If Desktop shows as running but the pipe is missing, quit Desktop fully (tray icon > Quit) and restart it; stale backend processes can wedge the pipe.
- Corrupt settings can prevent boot: resetting via Troubleshoot > Reset to factory defaults fixes most no-boot cases, at the cost of local images/containers.
- Antivirus or group policy blocking the backend service produces the same symptom; check the Desktop logs at %APPDATA%\Docker\log.