Managed Settings Builder

A dependency-free web UI that walks enterprise admins through creating GitHub Copilot's managed-settings.json.

Use

Live site: https://shakerg.github.io/enterprise-managed-settings/

Or open index.html locally in a browser (no server or build step needed).

1. Pick a starting point (Blank, Starter, Secure baseline, Locked down) or Import an existing file.
2. Step through Models, Permissions, Features & access (including an optional organization help page for Copilot CLI), Plugins, MCP servers, Sandbox, and Telemetry. Turn on Manage where available; unmanaged settings are omitted, while the help page is included when populated. The Default model pick list is loaded live from the Supported AI models docs page (cached for 24 hours, with a bundled snapshot in js/models-snapshot.js as an offline fallback). A custom model ID can also be entered. Network host allow/block rules are not supported in server-managed managed-settings.json; configure those through Copilot CLI user or device-managed settings.
3. Switch the live preview between generated JSON and a plain-English summary. Copy the summary as Markdown, or review contextual policy notes and validation before downloading or copying the JSON.
4. Commit it to copilot/managed-settings.json in your enterprise's .github-private repository.

Progress, theme (system / light / dark), and dismissed policy notes are saved in localStorage. Nothing is uploaded.

Structure

Path	Purpose
index.html	App shell
css/styles.css	Design tokens (black/charcoal palette, light & dark themes) and components
js/settings-core.js	Pure logic: state → JSON, JSON → state, validation, presets, model list parsing
js/models-snapshot.js	Offline fallback list of supported models
js/whats-new.js	Changelog behind the "What's new" banner and "New" badges
js/app.js	Wizard UI
tests/	Unit tests for settings-core.js

Tests

• Browser: open tests/index.html.
• Node: node --test tests/settings-core.test.js (also runs in CI on every push and PR)

Keeping it current

Two scheduled GitHub Actions workflows run daily:

Workflow	What it does
update-models.yml	Regenerates js/models-snapshot.js from the Supported AI models docs page, commits it if the list changed, adds a "What's new" entry, and redeploys Pages.
docs-watch.yml	Compares the managed settings reference with docs-baseline/enterprise-managed-settings.md. When it changes, opens a docs-change issue with the diff and assigns it to the Copilot coding agent, which opens a PR.

Both can also be run manually from the Actions tab.

What's new

Returning visitors see a dismissible What's new banner listing changes since their last visit, and recently changed steps and settings show a New badge for 30 days. Entries live in js/whats-new.js: the model workflow adds them automatically, and new or changed settings should add one by hand (the docs-watch issue asks Copilot to).

To let docs-watch.yml assign issues to Copilot, add a repository secret named COPILOT_ASSIGN_TOKEN containing a fine-grained personal access token for this repository with read and write access to Issues, Pull requests, Contents, and Actions, and read access to Metadata. Without it, the issue is still opened and you can assign Copilot manually.

To check the token after adding or rotating it, run gh workflow run docs-watch.yml -f check_token=true. This confirms Copilot can be assigned without opening an issue.