---
title: Introduction
description: Welcome to the MCPLambda Documentation.
---

## What is MCPLambda?

The artificial intelligence landscape is shifting towards **autonomous agents**. The Model Context Protocol (MCP) is the open standard that connects these AI models to external tools, databases, and APIs.

However, building agents is easy; managing the infrastructure to power them is hard. MCP servers are often stateful, requiring persistent connections and complex orchestration.

**MCPLambda** is a first-of-its-kind Platform-as-a-Service (PaaS) engineered to bridge this gap. It provides a "Vercel-like" experience for AI infrastructure, allowing you to deploy, manage, and scale MCP servers in seconds.

### Why MCPLambda?

- **Infrastructure-as-Code to Production-in-Seconds:** Deploy directly from package managers (`npx`, `uvx`), Git repositories, or Docker images.
- **Registry discovery:** Browse a curated Official + Community catalog (ToolHive-backed) and install into a project — without giving up bring-your-own-server.
- **Stateful by Design:** Seamlessly handle long-lived connections and persistent storage requirements without manual Kubernetes configuration.
- **Tool analytics:** Per-deployment call volume, error rates, p50/p95 tool latency, and a searchable invocation log.
- **CLI + agent control plane:** `mcpl` for terminals; service account tokens and the MCPLambda MCP server for AI-driven ops.
- **Security First:** Fine-grained tool management and secure secret injection.
- **Developer Velocity:** Abstract away the complexities of Kubernetes, networking, and storage so you can focus on building agents.

## Core Concepts

To get the most out of MCPLambda, it's helpful to understand a few key concepts:

- **Deployments:** A running instance of your MCP server.
- **Projects:** Logical groupings of deployments, secrets, and team members.
- **Registry:** Curated Official and Community MCP servers you can search and install; BYO package/Git/image remains first-class.
- **Tool analytics:** Per-deployment metrics for tool calls, errors, latency, and invocation history.
- **Service account tokens / MCP control plane:** `mcpl_sat_` tokens and the hosted MCPLambda MCP server (`api.mcplambda.io/mcp`) so agents and automation can manage infrastructure.
- **Compute Units:** A flexible resource model that scales with your needs (Small, Medium, and Large profiles).
- **Transport:** How your agent communicates with the server (Standard I/O, Streamable HTTP, or SSE).

## Next Steps

<Cards>
  <Card
    title="Quick Start"
    href="/docs/getting-started/"
    description="Deploy your first MCP server via dashboard or CLI."
  />
  <Card
    title="Deployment Strategies"
    href="/docs/deployment-strategies/"
    description="Learn about GitOps, Docker, and Package-based deployments."
  />
  <Card
    title="MCP Server Registry"
    href="/docs/registry/"
    description="Browse and install Official and Community MCP servers."
  />
  <Card
    title="Tool-Usage Analytics"
    href="/docs/analytics/"
    description="Call volume, error rates, latency, and invocation logs."
  />
  <Card
    title="The mcpl CLI"
    href="/docs/cli/"
    description="Manage your infrastructure directly from the terminal."
  />
  <Card
    title="MCPLambda MCP Server"
    href="/docs/mcp-server/"
    description="Manage your deployments directly from your AI client using the MCPLambda MCP server."
  />
  <Card
    title="Agent Guide"
    href="/docs/agents/"
    description="Instructions for AI agents to deploy MCP servers on behalf of users."
  />
  <Card
    title="Community Discord"
    href="https://discord.gg/58uWdd73tF"
    description="Join our Discord community for support and discussion."
  />
</Cards>