DocumentationGetting started

Getting started

One assistant. All your tools.

Validate your key, choose your clients and let XEDOC make the connection.

Before you run it

#
  • Node.js 20 or newer. Download the current LTS from nodejs.org.
  • An API key from your XEDOC workspace.
  • A current version of each client you want to connect. Codex command authentication is verified with Codex CLI 0.153.4.
  • Fully quit desktop clients before setup so they do not overwrite the new configuration when closing.

Run without installing

#

This downloads the versioned XEDOC package and starts the assistant. On Windows, npx.cmd also works when PowerShell blocks npx.ps1.

Windowspowershell
npx.cmd --yes https://xedoc.top/downloads/xedoc-setup-1.0.1.tgz
  1. Enter your XEDOC key

    Paste the key when asked. Input is hidden. Press Enter after pasting.

  2. Choose your clients

    Select one or more clients. The assistant fetches the current model catalog and prepares the appropriate configuration for each.

  3. Review and apply

    Review the destination files. Existing configurations are backed up before changes are applied.

  4. Reopen your client

    Choose the XEDOC provider or a model, then start a new conversation.

Keep the xedoc command

#
Windowspowershell
npm.cmd install -g https://xedoc.top/downloads/xedoc-setup-1.0.1.tgz --no-audit --no-fund
xedoc

No global install permission?

Use the run-once command above. It does not require installing the package globally. If a newly installed xedoc command is not found, open a new terminal and check that the npm global bin directory is on PATH. If PowerShell blocks xedoc.ps1, run xedoc.cmd.

What the assistant configures

#
ClientConnectionSelection
Codex CLIA separate XEDOC profile and model catalogStart with codex --profile xedoc
Codex AppThe shared Codex configuration, XEDOC catalog and private credential helperRestart the app and choose a model
Claude CodeXEDOC environment settings and model aliasesUse /model
Claude DesktopA named XEDOC inference configurationRestart and open the Code tab
OpenCodeAn XEDOC provider with the current model listUse /models

The assistant preserves unrelated settings. It never replaces Codex auth.json. Private credential files and backups stay on your computer; setup does not send them to XEDOC. Only the supplied key is sent to the XEDOC API to validate access and retrieve models.

Change a key or refresh models

#

Run xedoc again to validate a replacement key or refresh client model lists. Use the same selected clients. A repeated setup keeps the existing XEDOC connection instead of creating duplicate providers.

If a client configuration is invalid or changes while setup is running, setup stops rather than overwriting it. Backups are saved in the XEDOC configuration directory shown by the assistant. Linked configuration files require manual setup. Windows and WSL use separate homes; run setup in the environment where your client runs.

Prefer to configure things yourself?

#

Each integration guide also includes its connection settings and configuration format. Start with the guide for the client you use.