> ## Documentation Index
> Fetch the complete documentation index at: https://docs.morpheum.org/llms.txt
> Use this file to discover all available pages before exploring further.

# Deploy an agent portal

> Set up a Morpheum AgentPortal node for low-latency AI agents, automation, and agent-facing portal traffic

## Who this is for

Teams running **AI agents**, automation, or agent products that need optimized paths to chain data and portal features—without operating **consensus**. AgentPortal nodes tune behavior for **agent workloads** (latency, caching, and portal-oriented settings) rather than generic RPC alone.

***

## Before you start

* **Workload:** Clarify expected **QPS**, payload sizes, and whether traffic is authenticated or public.
* **Security:** Agent traffic often carries signatures and payments; apply **TLS**, **authentication**, and **rate limits** like any public API.
* **Relationship to validators:** This role does **not** replace a validator. It participates in the network as a follower node with portal-focused configuration.

***

## Initialize the node

1. Install the Morpheum CLI.
2. Run **init** with role **agent-portal** (or the exact role name shown in the **init** subcommand’s help for your release).
3. Set **data directory**, **moniker**, and **chain ID**.

Post-init layouts may include **agent-related directories or templates**—follow any summary printed at the end of init for next steps.

***

## Configuration

Review:

* **Agent / portal** — Features that enable or tune portal behavior, shard affinity, or caching (names vary by release; use your generated config as the source of truth).
* **API exposure** — Bind addresses suitable for your agents (internal network vs internet).
* **Peers** — Good connectivity to stay near the chain tip for responsive agent actions.

***

## Run the node

Start with **run** (non-validator). Monitor **latency**, **error rates**, and **cache hit rates** if your stack exposes them.

Co-locate with agents or place in the same region as your agent fleet to meet latency goals.

***

## Related commands

The CLI may expose an **agent** subcommand for registration and agent-specific workflows. See [Node CLI](/node/cli) for what your installed version supports.

***

## Related

* [Node CLI](/node/cli)
* [Deploy a full node](/node/deploy-full) — If you need a general operator node without portal specialization