> ## Documentation Index > Fetch the complete documentation index at: https://docs.getunbound.ai/llms.txt > Use this file to discover all available pages before exploring further. # First-Run Setup > The onboarding wizard: naming your organization, installing Unbound, and turning on your first policies | | | | ----------- | ---------------------------------------------------------------------- | | **URL** | `/onboarding/setup`, then `/onboarding/configure-workspaces` | | **Sidebar** | None — setup runs full screen, with its own **First-Run Setup** header | | **Roles** | The first person to sign in for your organization | First-Run Setup is the flow every new organization meets. It names your organization, gets Unbound installed on a machine, records how your coding agents are licensed, and finishes by turning on your first policies org-wide. Until it is finished, opening any dashboard page returns you here, and `/onboarding` lands you on whichever part you still have to do. Once it is finished, `/onboarding` sends you to [Home](/dashboard/home). ## What's on the page ### Two parts | Part | URL | What it does | | ------------------------- | ---------------------------------- | ------------------------------------------------------------------------------------------------------------ | | **Create Organization** | `/onboarding/setup` | Names your organization. The field is pre-filled from your email domain, except for personal email providers | | **Configure Preferences** | `/onboarding/configure-workspaces` | The wizard below | ### The wizard steps The wizard is **seven to nine steps**: the plan steps appear only for the agents you pick, so choosing both Claude Code and Cursor gives you nine and choosing neither gives you seven. The header carries a progress track, and a scanning panel sits top right from the first step onward. | Step | What you choose | | ----------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Deployment** | **Install via CLI** for one or two machines by hand, or **Deploy via MDM** for a fleet — then your MDM: Microsoft Intune, Rippling, JumpCloud, Jamf, Iru, or **Other / custom** | | **Configure** | What the CLI does on the machine — the four capabilities below | | **Run & Complete** | Nothing to choose. The page shows the install command built from your answers, with your API key masked on screen and complete when you copy it. MDM deployments get platform tabs — **macOS / Linux** and **Windows** when discovery is the only capability, and **macOS**, **Linux** and **Windows** once **View & secure agent actions** is on as well. Copy from the tab that matches the machines you are deploying to. The scanning panel fills in as the machine reports, and **Continue** opens once at least one device has been scanned | | **Agent inventory** | Which coding agents your organization supports: Claude Code, Cursor, Codex, Gemini CLI, GitHub Copilot, Roo Code, Cline, Kilo Code, Open Code and Junie. You can refine the list later | | **Plans · Claude Code** | Every Claude Code plan your team is on, split into **Personal** and **Organization** tiers. Appears only if you picked Claude Code | | **Plans · Cursor** | The same for Cursor, and only if you picked Cursor | | **MCP gateway** | Whether you already run an MCP gateway. **Yes** asks for its URL | | **Priorities** | Rank what concerns you most — skill sprawl, MCP chaos, risky terminal ops, cost versus usage, and an audit trail for agentic actions | | **Load Policies** | Your first policy set, below | #### What the CLI does | Capability | | | ---------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Discover your coding agents** | Finds every AI tool on the machine, and the MCP servers and skills they are wired into · **Required**, always on | | **View & secure agent actions** | Installs the Unbound hooks, so policies are enforced on tool calls, file writes and shell commands as agents run · **Recommended** | | **Seed events from past activity** | Fills the dashboard with what the machine's agents did before the install. Claude Code, Codex and Copilot only, and offered once hooks are on · **Recommended** | | **Keep governance up to date** | Keeps the machine's setup in place and its discovered inventory current, checked daily. Offered on CLI installs only — picking **Deploy via MDM** removes it — and only where your organization has it turned on · **Optional** | ### Load Policies The last step turns on your baseline, in three parts: | Part | What happens | | ------------ | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | | **Intro** | Tells you what is coming — one question per category, about thirty seconds | | **The deck** | One card per category, seven in all, each asking a plain yes/no question. Keep a category and Unbound will create its policies; skip one and it creates none of them | | **Review** | One card per category you kept, with its policy count and its **Block** / **Audit** split. **Edit** expands a card to every policy in it, each with a **Block** / **Audit** toggle you can flip before you confirm | The seven categories, in the order the deck asks them: * **Production Systems** — *Can your agents deploy or change production directly?* * **Cloud & Infrastructure** — *Do your agents have cloud credentials (AWS, GCP, Azure, Terraform)?* * **Identity, Access & Privilege** — *Can your agents change access or escalate to root?* * **Secrets & Credentials** — *Should agents be able to read your secrets and tokens?* * **Databases & Data Movement** — *Do your agents have database access or move data in bulk?* * **Source Control & Delivery** — *Should agents be able to force-push or push to protected branches?* * **System & Runtime Safety** — *Should agents be able to run destructive commands like `rm -rf`?* [Recommended Starting Policies](/playbook/recommended-starting-policies) lists every policy in the pack, with the reasoning behind each default. The final button reads **Finish setup** when you kept at least one category and **Finish** when you kept none. Either way it creates the policies you kept and lands you on the dashboard, with a message telling you how many are live. **Prefer to do this later? Skip for now** finishes setup without creating any — you can author the same rules afterwards from [Policies — Agentic Use](/dashboard/policies/agentic-use). ## Filters and controls The wizard is built for the keyboard. | Key | What it does | | -------------- | -------------------------------------------------------------------------------------------------------------------------- | | **1** – **9** | Picks or toggles the numbered choice on the current step | | **Arrow keys** | Move focus between the choices, wrapping at either end — except on **Priorities**, where they stop at the ends | | **Enter** | Continues, once the step has what it needs | | **S** / **M** | Chooses CLI or MDM on the **Deployment** step, where **1** – **6** then pick the MDM | | **Y** / **N** | Answers the **MCP gateway** question, and keeps or skips a category on the policy deck — where **→** and **←** do the same | | **⌫** / **R** | Clears one ranking, or resets them all, on **Priorities** | Cards can also be clicked, and the policy deck can be swiped. **Back** returns to the previous step on every step after the first, and **Redo the questions** on the review sends you through the deck again. ## What you can do | Action | Result | | --------------------------------------------------------- | ---------------------------------------------------------------------------------- | | **Copy** on the install command | Copies the command with your real API key in place of the mask | | **Continue** | Moves to the next step. On **Run & Complete** it waits until a device has reported | | **Finish setup** — **Finish** when you kept no categories | Creates the policies you kept and opens the dashboard | | **Prefer to do this later? Skip for now** | Completes setup with no policies created | | **Redo the questions** | Returns to the category deck from the review | ## Messages you may see | Message | What it means | | ---------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------ | | `Please enter an organization name` | The organization name is required | | `Failed to create organization` | The organization was not created. Your entry stays on screen, so submitting again is enough | | `Failed to load pricing plans from server. Using default plans.` | The plan list could not be fetched, so the standard plans are shown. You can still choose and continue | | `Loading policy catalog…` | The policy pack is being fetched | | `Couldn't load the policy catalog` | The pack could not be fetched. Retry, or skip and apply it later from Policies | | `No default policies available` | Your organization has no default pack to apply | | `Couldn't load your policies` | The policies were not created, and setup stays open so you can try again | | `Setup complete` | Setup finished with no policies created | | `N policies are live` | Setup finished and the policies you kept are enforcing | ## Common questions **Can I skip setup and come back?** Setup has to finish before the dashboard opens — until it does, every page returns you here. Once it is done, `/onboarding` sends you to Home. **I skipped the policy questions. Can I still get the pack?** The deck runs once, at setup. Afterwards you create the policies you want from **Policies → Agentic Use → Create Policy** — [Recommended Starting Policies](/playbook/recommended-starting-policies) gives the exact values for each one. Where your organization has **Policy Packs**, that applies a curated starter set in one step. **Why don't I see the plan steps?** They appear only for the agents you picked. Pick Claude Code or Cursor on **Agent inventory** and the matching plan step appears. **Why is Continue greyed out on Run & Complete?** It opens once at least one device has been scanned. Run the install command, and the scanning panel will show the machine reporting. **Can I change these answers later?** Yes. Agents and devices live in [Discovery](/dashboard/discovery/summary), policies in [Policies — Agentic Use](/dashboard/policies/agentic-use), and fleet deployment in [Configure](/dashboard/configure). ## Related | Page | For | | ------------------------------------------------------------------------ | ------------------------------------------------------------------- | | [Recommended Starting Policies](/playbook/recommended-starting-policies) | Every policy in the pack the last step applies | | [Configure](/dashboard/configure) | Setting up another coding tool, and finding your API key afterwards | | [Unbound CLI](/cli/overview) | What the install command sets up, and how to re-run it | | [Discovery — Summary](/dashboard/discovery/summary) | Confirming devices are reporting | | [Policies — Agentic Use](/dashboard/policies/agentic-use) | Editing, rescoping or removing what setup turned on |