Use a docly CLI working copy instead of git

Misconception By design
Docly has no git service: there is no git remote to clone from or push to. A Docly workspace mapped as a drive is the live application, not a working copy, and running git init inside it creates a .git directory in the running site on a slow network filesystem. When you want a local working copy, a diff and a review step before anything goes live, use the docly CLI: docly clone gives you local files, docly status and docly diff show what you changed, and docly push publishes it.
Applies to: Workspace setupAI agents

What you'll see

Someone — often an AI coding agent on its first pass — opens a mapped Docly drive, finds no .git directory, and runs git init to get version control before starting work. Or they look for a git remote to clone the site from. Neither exists, and the first one damages the live site.

What's actually happening

Docly has no git service. There is no git remote for a workspace or a folder, and no git URL to clone. Version control of a Docly site is not done with git.

The mapped drive is the live application. It is a WebDAV mount, so there is no working copy there — the files are the running site. Running git init against it does not give you version control of your deployment: it creates a .git directory inside the application, on a filesystem where every read and write is an HTTP round trip. The result is a repository that is slow to the point of unusable, sitting in a folder your users can see, tracking a tree that changes underneath it whenever anyone edits through the Docly UI.

The working copy you are looking for is the docly CLI. docly clone downloads a folder into local files, docly status and docly diff show what you changed against the server, and nothing goes live until docly push. Writing is locked until someone opens one folder for writing, so the blast radius is bounded. Documents whose schema keeps versions also have server-side history (docly history).

A CLI working copy is ordinary local files. If you want your own git history on top of it, you can run git init in the working copy on your own disk — that is a private repository for you, not a connection to Docly, and it is never pushed anywhere by Docly.

What to do

On a mapped drive: never run git init. If you need a copy before editing there, copy the files out first — see Work safely on a live site.

For a working copy, use the CLI:

docly login
docly clone <folder> work && cd work
# edit the files
docly status
docly diff
docly unlock <folder>   # writing is locked until a folder is opened
docly push

Run docly with no arguments for the full command reference. For AI agents, see How an AI agent should use the docly CLI.

Choosing between the drive and the CLI

Mapped drivedocly CLI
Change goes liveOn save, immediatelyOn docly push
HistoryServer-side versions for documents whose schema keeps them; none for code filesThe same server-side versions through docly history; code files (.hash, .js, .css) have none
Review before publishNot possibledocly status and docly diff
SpeedEvery operation is an HTTP round tripLocal disk after one clone; server-side search instead of walking a tree
Write protectionNoneLocked by default; one folder opened at a time
Good forQuick edits, throwaway pages, low-risk sitesScripts and AI agents, batch work, anything where you want a diff before it is live

State which one a workspace uses in its agent instruction file, so an agent does not have to guess — see Set up AI agent instruction files for a workspace.