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

# Setup

> Clone the template and run it locally in five minutes.

This page gets the Agent Platform template running on your machine: a FastAPI service on port 8000, a Postgres for sessions and memory, and two reference agents (`code_search`, `web_search`).

## Prerequisites

* Docker Desktop ([install](https://docs.docker.com/desktop/))
* OpenAI API key ([get one](https://platform.openai.com/api-keys))

## Run locally

<Steps>
  <Step title="Clone the template">
    ```bash theme={null}
    git clone https://github.com/agno-agi/agentos-railway-template.git agent-platform
    cd agent-platform
    ```

    The remote stays at `agentos-railway-template`; the local directory is `agent-platform` because that's the platform you're building.
  </Step>

  <Step title="Configure your environment">
    ```bash theme={null}
    cp example.env .env
    ```

    Open `.env` and set `OPENAI_API_KEY`. Everything else has sensible defaults.
  </Step>

  <Step title="Start the platform">
    ```bash theme={null}
    docker compose up -d --build
    ```

    Two containers come up: a FastAPI app and a Postgres database. First run pulls the base image and builds the container. Subsequent starts are instant.
  </Step>

  <Step title="Verify it's running">
    ```bash theme={null}
    curl http://localhost:8000/health
    ```

    Expect `{"status":"ok"}`. If you get connection refused, the container is still starting; check `docker compose logs -f` and wait for the Agno banner.

    The OpenAPI spec is at [http://localhost:8000/docs](http://localhost:8000/docs).
  </Step>
</Steps>

## Connect the AgentOS UI

<Snippet file="connect-agent-os-ui.mdx" />

You should see two agents in the sidebar:

| Agent            | What it does                                                           |
| ---------------- | ---------------------------------------------------------------------- |
| **code\_search** | Reads files in the workspace and answers questions about the codebase. |
| **web\_search**  | Searches the web and synthesizes answers grounded in citations.        |

Try one prompt against each:

> *"What's in this repo?"* → `code_search` lists the directory and explains the structure.

> *"What did Anthropic publish about agents recently?"* → `web_search` returns a summary with URLs.

## What you have now

* A FastAPI runtime on port 8000 with sessions, scheduling, and 50+ endpoints.
* A Postgres database holding sessions, memory, and traces.
* Two reference agents you can replace, extend, or use as patterns.
* A `docs/` folder (inside the cloned repo) of Claude Code prompts that build new agents for you.

## Next

[Create an agent →](/tutorials/agent-platform/create-an-agent)