---
title: Getting Started
description: Set up your account and deploy your first server.
---

Get from zero to a live MCP endpoint in a few minutes. Pick the path that fits how you work — dashboard UI or the `mcpl` CLI — then connect your AI client.

---

## 1. Sign In

Visit [mcplambda.io](https://mcplambda.io) and sign in with **GitHub** or **Google**. That creates your account and first project.

---

## Path A: Dashboard

1. Click **Create Deployment**.
2. Choose a **Deployment Flow** — Package, Git, or Image — **or** open **Registry** in the sidebar, find a server, and install it (pre-fills the form).
3. Example package source: `npx://@mcp/server-time`.
4. Name the deployment (e.g. `time-server`).
5. Click **Deploy**.

Within seconds you should see live status and logs. When the status is running, copy the **Deployment URL**.

---

## Path B: mcpl CLI

### Install

**macOS / Linux:**

```bash
curl -fsSL https://mcplambda.io/mcpl/install.sh | sh
```

**Windows (PowerShell 5.1+):**

```powershell
irm https://mcplambda.io/mcpl/install.ps1 | iex
```

### Login and deploy

```bash
mcpl login
mcpl deploy npx://@mcp/server-time
```

`mcpl login` opens the browser for GitHub or Google. Tokens are stored in `~/.mcpl/config.json`. The deploy command streams build logs and prints the deployment URL when the server is running.

Optional: discover a server first, then deploy the package or image it lists:

```bash
mcpl registry search time
mcpl registry info <name>
mcpl deploy npx://@mcp/server-time
```

Full command reference: [The mcpl CLI](/docs/cli/).

---

## Registry as an alternative install path

Prefer browsing over typing package names? Use the [MCP Server Registry](/docs/registry/):

- **Dashboard:** Sidebar → **Registry** → search / filter → install into your project.
- **CLI:** `mcpl registry search` / `mcpl registry info`.
- **API:** public `GET /v1/registry/servers` (no auth for reads).
- **AI client:** `search_registry` via the [MCPLambda MCP server](/docs/mcp-server/).

Bring-your-own package, Git repo, or Docker image remains fully supported — the registry is convenience discovery, not a requirement.

---

## 2. Connect Your AI Client

Use the **Deployment URL** from the dashboard or CLI output.

- **Cursor / VS Code extensions:** paste the URL in MCP settings (Streamable HTTP or SSE, depending on transport).
- **Claude Desktop and other local clients:** use the URL (and API key if you chose `auth_type: key`; default `oauth` is handled by the client).

Step-by-step client configs: [Connecting AI Clients](/docs/ai-clients/).

---

## 3. What to do next

Once traffic hits the server, open the deployment → **Analytics** tab for tool call volume, error rates, and latency ([Tool-Usage Analytics](/docs/analytics/)).

<Cards>
  <Card
    title="Deployment Strategies"
    href="/docs/deployment-strategies/"
    description="Package, Git, Docker, and registry discovery."
  />
  <Card
    title="MCP Server Registry"
    href="/docs/registry/"
    description="Browse Official and Community servers."
  />
  <Card
    title="The mcpl CLI"
    href="/docs/cli/"
    description="Full terminal reference."
  />
  <Card
    title="MCPLambda MCP Server"
    href="/docs/mcp-server/"
    description="Manage deployments from your AI client."
  />
</Cards>