touchups & add MCP Gateway feature

Signed-off-by: Craig <[email protected]>

Author: craig-osterhout

content/guides/agentic-ai.md

@@ -45,6 +45,7 @@ shows how Docker ties them all together with the following tools:
   Toolkit](../manuals/ai/mcp-catalog-and-toolkit/_index.md) helps you discover
   and securely run external tools, like APIs and databases, using the Model
   Context Protocol (MCP).
+- [Docker MCP Gateway](/manuals/ai/mcp-gateway/_index.md) lets you orchestrate and manage MCP servers.
 - [Docker Cloud](/cloud/) provides a powerful, GPU-accelerated
   environment to run your AI applications with the same Compose-based
   workflow you use locally.
@@ -112,18 +113,18 @@ To run the application in Docker Cloud, follow these steps:
    and run it.
 
 4. Visit [http://localhost:8080](http://localhost:8080). Enter something in the
-prompt and hit enter. An agent searches DuckDuckGo and another agent revises the
-output.
+   prompt and hit enter. An agent searches DuckDuckGo and another agent revises
+   the output.
 
-  ![Screenshot of the application](./images/agentic-ai-app.png)
+   ![Screenshot of the application](./images/agentic-ai-app.png)
 
 5. Press ctrl-c in the terminal to stop the application when you're done.
 
 6. Run the following command to stop Docker Cloud:
 
-```console
-$ docker cloud stop
-```
+   ```console
+   $ docker cloud stop
+   ```
 
 ## Step 3: Optional. Run the application locally
 
@@ -175,37 +176,40 @@ instructions below to help you understand each line.
 services:
   adk:
     build:
-      context: .                          # Build from the current directory
-      dockerfile: Dockerfile              # Use the specified Dockerfile
+      context: .
     ports:
-      - "8080:8080"                       # Map host port 8080 to container port 8080
+      # expose port for web interface
+      - "8080:8080"
     environment:
-      - MCPGATEWAY_ENDPOINT=http://mcp-gateway:8811/sse  # Tell the app where to find the MCP gateway
-    command: ["web"]                      # Start the ADK web interface
-    extra_hosts:
-      - host.docker.internal:host-gateway # Allow container to reach the host machine
+      # point adk at the MCP gateway
+      - MCPGATEWAY_ENDPOINT=http://mcp-gateway:8811/sse
     depends_on:
-      - docker-model-runner              # Ensure model service starts before this
-      - mcp-gateway                      # Ensure gateway service starts before this
+      # model_runner provider starts first then injects environment variables
+      # MODEL_RUNNER_MODEL name
+      # MODEL_RUNNER_URL OpenAI compatible API endpoint
+      - model_runner
 
-  docker-model-runner:
+  model_runner:
     provider:
-      type: model                        # Use Docker Model Runner to serve a model
+      type: model
       options:
-        model: ai/gemma3-qat:27B-Q4_K_M  # Specify the model to run (27B Gemma3 QAT)
-    environment:
-      - LLAMA_ARGS=--ctx-size 20000 --jinja -ngl 100  # Runtime options for the model engine
+        # pre-pull the model when starting Docker Model Runner
+        model: ai/gemma3-qat:27B-Q4_K_M
+        # increase context size to handle search results
+        context-size: 20000
 
   mcp-gateway:
-    image: docker/agents_gateway:v2       # Run Docker's MCP gateway service
+    # agents_gateway secures your MCP servers
+    image: docker/agents_gateway:v2
     ports:
-      - "8811:8811"                       # Expose the gateway on port 8811
+      - "8811:8811"
     command:
-      - --transport=sse                  # Use SSE (Server-Sent Events) for agent communication
-      - --servers=duckduckgo             # Enable DuckDuckGo tool server for web search
+      - --transport=sse
+      # add any MCP servers you want to use
+      - --servers=duckduckgo
     volumes:
-      - /var/run/docker.sock:/var/run/docker.sock  # Allow gateway to interact with Docker on host
-
+      # mount docker socket to run MCP containers
+      - /var/run/docker.sock:/var/run/docker.sock
 ```
 
 The first notable element here is the `provider` section that specifies `type:
@@ -221,10 +225,11 @@ The `options` section defines the specific model to run, in this case,
 The second notable element is `image: docker/agents_gateway:v2`, which indicates
 that the MCP gateway service will use the [docker/agents_gateway:v2
 image](https://hub.docker.com/r/docker/agents_gateway). This image is Docker's
-[open source MCP gateway](https://github.com/docker/docker-mcp/) that enables your application to connect to MCP servers,
-which expose tools that models can call. In this example, it uses the
-[`duckduckgo` MCP server](https://hub.docker.com/mcp/server/duckduckgo/overview)
-to perform web searches.
+open source [MCP Gateway](https://github.com/docker/docker-mcp/) that enables
+your application to connect to MCP servers, which expose tools that models can
+call. In this example, it uses the [`duckduckgo` MCP
+server](https://hub.docker.com/mcp/server/duckduckgo/overview) to perform web
+searches.
 
 >  [!TIP]
 >
@@ -237,13 +242,6 @@ external tools and services using the Model Context Protocol.
 
 ## Step 5: Review the application
 
-The `adk` web application is an agent implementation that connects to the MCP
-gateway and a local model through environment variables and API calls. It uses
-the [ADK (Agent Development Kit)](https://github.com/google/adk-python) to
-define two agents that evaluate responses, verify claims, and revise answers.
-These agents load a language model (served via Docker Model Runner) and connect
-to external tools using the Model Context Protocol (MCP).
-
 The `adk` web application is an agent implementation that connects to the MCP
 gateway and a local model through environment variables and API calls. It uses
 the [ADK (Agent Development Kit)](https://github.com/google/adk-python) to
@@ -301,6 +299,11 @@ suite of tools:
 - [Docker MCP Catalog and
   Toolkit](../manuals/ai/mcp-catalog-and-toolkit/_index.md): Launch and manage
   tool integrations that follow the Model Context Protocol (MCP) standard.
+- [Docker MCP Gateway](/manuals/ai/mcp-gateway/_index.md): Orchestrate and manage
+  MCP servers to connect agents to external tools and services.
+- [Docker Compose](../manuals/compose/_index.md): Define and run multi-container
+  applications with a single file, using the same workflow locally and in the
+  cloud.
 - [Docker Cloud](/cloud/): Deploy and scale your application
   in a secure, managed cloud environment.