Enshrouded server password: answers on saves and settings

The simplest way is to stop the server, open enshrouded_server.json and set the password for a user group, because in current versions passwords and permissions are set per user group in this file. Let the agent make the edit with edit_file, read the diff and accept it in Ask first. Then start the server and join. Never post a file that holds a real password.

The world is stored in the folder named by saveDirectory in enshrouded_server.json. Open the file with read_file to see the path, or let list_dir show the folder, so that you know exactly what a backup has to copy. Check the path before every copy, because a changed setting changes the folder too, and keep the copy outside the server folder, ideally on another drive.

The simplest way is to stop the server and copy the saveDirectory folder with Robocopy through Shell before every update, settings change and experiment. The card shows the real exit code of the copy. Then test that you can restore it. A copy that you never restored is only a hope, and no copy is a promise.

The server rewrites the file with its defaults when a field is missing, which is one known cause. Edit the file only while the server is stopped, copy it before each change, and ask the agent to compare the copy with the current file using read_file and grep. A wrong comma or quote makes the file invalid too.

Rubber banding is the effect when a player's movement snaps back, and no single setting explains it. Read the server log with grep for the first error, check the settings file and the ports, and compare your hardware with the official server requirements. Server stability with more than three players depends on the world and the hardware, so change one thing at a time.

Yes, buildlys.com is the official VarintKeel website. The download button gives the original setup program of about 34 MB, and you can compare its SHA-256 checksum with the published manifest before you run it. Check the checksum first, on every download. Questions about the app, and reports of problems, always go through the Send feedback button inside the app.

No, the agent runs tools and shows the real result, but nothing guarantees that a change is correct. That is why edit_file shows a diff before you accept it in Ask first, and why you start the server afterwards and read the log with grep. Test every change yourself, on a copy of the folder first, and treat a quiet log as a hint, not as proof.

Ask first is the mode to start with: it pauses before actions that change something, and the diff waits for your OK. Plan only writes a plan and runs nothing, which suits a first look. Auto-run acts at once, so keep it for work on a copy. You can switch modes at any moment, for example from Plan only to Ask first once you have read the plan.

The tool grep searches the log file for the word you give it, and read_file shows the lines around the match. Ask the agent to search for the first error line, then to read its context. The agent reports what it found; you decide whether the fix is right and test it by starting the server again.

Yes, if you own or rent the server and may administer it, because Shell can run in the chosen environment, including SSH to a server you own. The file tools work in the chat's working folder. Keep the agent away from servers that belong to other people, because the agent does not enforce that, and ask the host first if the machine is shared.

The model endpoint you pick decides it. With a cloud model, your prompt and the file text the agent reads go to that provider; with a local model nothing leaves your PC. Your chats, working folders and API keys stay on the PC, and secrets in tool output are redacted before they reach the model.

No, VarintKeel runs locally and sends no telemetry. Prompts go only to the model endpoint that you choose, and the optional network features are web search, control of chats from a phone browser and update checks. The privacy page lists all of them, together with what this website records and what the app keeps on your PC.

The simplest way is to run certutil -hashfile on the downloaded file in cmd.exe and compare the SHA-256 value with the published manifest. The installer is not code-signed yet, so Windows SmartScreen may warn; choose More info and then Run anyway only after the checksum matches. If the values differ, do not run the file.

The Send feedback button inside the app is the way to report a problem or to ask a question. Describe what you asked the agent to do, what you expected and what the card showed, and add the name of the file or command involved. A concrete report is easier to repeat, because it says which tool and which mode were involved.