Skip to content

[WIP] (docs) make landing MCP doc root more actionable #402

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
wants to merge 1 commit into from
Closed

[WIP] (docs) make landing MCP doc root more actionable #402

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
127 changes: 127 additions & 0 deletions docs/source/how-it-works.mdx
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,127 @@
---
title: How it Works
subtitle: Introduction to MCP and how the Apollo MCP Server works

Check notice on line 3 in docs/source/how-it-works.mdx

View check run for this annotation

Apollo Librarian / AI Style Review

docs/source/how-it-works.mdx#L3 

For conceptual pages, subtitles should start with 'Learn'.

```suggestion
subtitle: Learn about MCP and how the Apollo MCP Server works
```
---

Apollo MCP Server provides a standard way for AI models to access and orchestrate your GraphQL APIs.

## What is MCP?

Check warning on line 8 in docs/source/how-it-works.mdx

View check run for this annotation

Apollo Librarian / AI Style Review

docs/source/how-it-works.mdx#L8 

For conceptual documents, headings should use gerunds for consistency.

```suggestion
## Understanding MCP
```

[Model Context Protocol (MCP)](https://modelcontextprotocol.io/introduction) is an open protocol that standardizes how applications provide context to AI models like Large Language Models (LLM). MCP enables LLMs and AI agents to indirectly fetch data from external sources.

MCP follows a client-server architecture. MCP servers expose functions, called _tools_, that MCP clients can invoke.

## What is Apollo MCP Server?

Check warning on line 14 in docs/source/how-it-works.mdx

View check run for this annotation

Apollo Librarian / AI Style Review

docs/source/how-it-works.mdx#L14 

For conceptual documents, headings should use gerunds for consistency.

```suggestion
## Understanding Apollo MCP Server
```

Apollo MCP Server is an implementation of an MCP server. It makes GraphQL API operations available to AI clients as MCP tools. You can use Apollo MCP Server with any GraphQL API.

The GraphQL operations can be configured from persisted queries, which are predefined, approved lists of operations that are registered with and maintained by a graph. The operations can also be determined by AI introspecting your graph schema.

Check warning on line 18 in docs/source/how-it-works.mdx

View check run for this annotation

Apollo Librarian / AI Style Review

docs/source/how-it-works.mdx#L18 

Adopt an opinionated voice by prescribing a recommended path instead of just listing options.

```suggestion
The recommended approach is to configure GraphQL operations from persisted queries, which are predefined, approved lists of operations registered with your graph. You can also enable AI introspection of your graph schema to determine operations.
```

Apollo MCP Server is deployable in local environments via Apollo's Rover CLI or in containerized services in your cloud infrastructure. It can expose an MCP endpoint using Streamable HTTP for communication with AI clients.

Check notice on line 20 in docs/source/how-it-works.mdx

View check run for this annotation

Apollo Librarian / AI Style Review

docs/source/how-it-works.mdx#L20 

Use the imperative mood and active voice for clearer, more direct instructions.

```suggestion
Deploy Apollo MCP Server in local environments with the Rover CLI or as a containerized service in your cloud infrastructure. It exposes an MCP endpoint using Streamable HTTP for communication with AI clients.
```

## How Apollo MCP Server works

Check warning on line 22 in docs/source/how-it-works.mdx

View check run for this annotation

Apollo Librarian / AI Style Review

docs/source/how-it-works.mdx#L22 

For conceptual documents, headings should use gerunds for consistency.

```suggestion
## Understanding how Apollo MCP Server works
```

Apollo MCP Server bridges AI applications and your GraphQL APIs, translating GraphQL operations into MCP tools that AI models can discover and use.

```mermaid
graph LR
%% Nodes
AI["AI Application\n(Claude, ChatGPT, etc.)"]
MCPClient["MCP Client\n(Built into AI app)"]
MCPServer["Apollo MCP Server"]
GraphQL["GraphQL API\n(Your Graph)"]
Data["Your Data Sources\n(Databases, APIs, etc.)"]

%% Connections
AI <-->|"Natural Language\nRequests"| MCPClient
MCPClient <-->|"MCP Protocol\n(stdio/Streamable HTTP)"| MCPServer
MCPServer <-->|"GraphQL\nOperations"| GraphQL
GraphQL <-->|"Data\nQueries"| Data

%% Tool Generation
subgraph ToolGeneration[Tool Generation]
direction TB
OpFiles["Operation Files\n(.graphql)"]
PQM["Persisted Query\nManifests"]
Introspection["Schema\nIntrospection"]
Tools["MCP Tools"]

OpFiles --> Tools
PQM --> Tools
Introspection --> Tools
end

MCPServer -.->|"Exposes"| Tools
Tools -.->|"Available to"| MCPClient

%% Styling
classDef default stroke-width:1px
classDef aiClient stroke-width:2px
classDef mcpComponent stroke-width:2px
classDef apolloComponent stroke-width:2px
classDef apiComponent stroke-width:2px
classDef dataComponent stroke-width:2px

class AI aiClient
class MCPClient mcpComponent
class MCPServer apolloComponent
class GraphQL apiComponent
class Data dataComponent
class OpFiles,PQM,Introspection apolloComponent
class Tools mcpComponent
```

The architecture enables intelligent API orchestration through these components:

* AI Applications: Tools like Claude Desktop or ChatGPT connect to Apollo MCP Server through their built-in MCP clients, making requests in natural language.
* Transport Options: Communication happens over stdio for local development or Streamable HTTP.

Check notice on line 77 in docs/source/how-it-works.mdx

View check run for this annotation

Apollo Librarian / AI Style Review

docs/source/how-it-works.mdx#L77 

Use code font for CLI-related terms like `stdio`.

```suggestion
* Transport Options: Communication happens over <code>stdio</code> for local development or Streamable HTTP. 
```
* Tool Generation: Apollo MCP Server creates MCP tools from your GraphQL operations using:
* Operation Files: Individual `.graphql` files for specific queries or mutations
* Persisted Query Manifests: Pre-approved operation lists from Apollo GraphOS
* Schema Introspection: Dynamic operation discovery for flexible AI exploration

* Secure Execution: When invoked, the server executes GraphQL operations against your API endpoint, respecting all existing authentication, headers, and security policies.
* Existing Infrastructure: Your GraphQL API handles requests normally, with Apollo MCP Server acting as a controlled gateway rather than requiring any changes to your graph.

This design lets you expose precise GraphQL capabilities to AI while maintaining complete control over data access and security.

### Example usage

Check warning on line 88 in docs/source/how-it-works.mdx

View check run for this annotation

Apollo Librarian / AI Style Review

docs/source/how-it-works.mdx#L88 

For conceptual documents, headings should use gerunds for consistency.

```suggestion
### Reviewing an example
```

Once configured, AI applications can use your GraphQL operations naturally:

Check notice on line 90 in docs/source/how-it-works.mdx

View check run for this annotation

Apollo Librarian / AI Style Review

docs/source/how-it-works.mdx#L90 

Replace 'once' with 'after' when describing a sequence of events for better clarity.

```suggestion
After configuration, AI applications can use your GraphQL operations naturally:
```

> User: "Show me the astronauts currently in space"
>
> Claude: *Uses GetAstronautsCurrentlyInSpace tool to query your GraphQL API*

Check notice on line 94 in docs/source/how-it-works.mdx

View check run for this annotation

Apollo Librarian / AI Style Review

docs/source/how-it-works.mdx#L94 

Do not use italics for emphasis. Use code font for symbols that appear in code, such as tool names.

```suggestion
> Claude: Uses the <code>GetAstronautsCurrentlyInSpace</code> tool to query your GraphQL API
```
>
> "There are currently 7 astronauts aboard the ISS..."

## Why GraphQL for AI?

Check warning on line 98 in docs/source/how-it-works.mdx

View check run for this annotation

Apollo Librarian / AI Style Review

docs/source/how-it-works.mdx#L98 

For conceptual documents, headings should use gerunds for consistency.

```suggestion
## Using GraphQL for AI
```

GraphQL's architecture provides unique advantages for AI-powered API orchestration:

**🎯 Deterministic Execution**: GraphQL's built-in relationship handling and query structure eliminate guesswork for AI models. The graph defines clear paths between data types, ensuring AI agents execute operations in the correct sequence without complex prompt engineering or error-prone orchestration logic.

Check warning on line 102 in docs/source/how-it-works.mdx

View check run for this annotation

Apollo Librarian / AI Style Review

docs/source/how-it-works.mdx#L102 

Remove the emoji, as it is not in the approved set. Remove bolding, as it should not be used for emphasis.

```suggestion
Deterministic execution: GraphQL's built-in relationship handling and query structure eliminate guesswork for AI models. The graph defines clear paths between data types, ensuring AI agents execute operations in the correct sequence without complex prompt engineering or error-prone orchestration logic.
```

**🛡️ Policy Enforcement**: Security policies and access controls apply consistently across all services within a single GraphQL query context. This unified enforcement model ensures AI operations respect organizational boundaries, even when spanning multiple underlying APIs or microservices.

Check warning on line 104 in docs/source/how-it-works.mdx

View check run for this annotation

Apollo Librarian / AI Style Review

docs/source/how-it-works.mdx#L104 

Remove the emoji, as it is not in the approved set. Remove bolding, as it should not be used for emphasis.

```suggestion
Policy enforcement: Security policies and access controls apply consistently across all services within a single GraphQL query context. This unified enforcement model ensures AI operations respect organizational boundaries, even when spanning multiple underlying APIs or microservices.
```

**⚡ Efficiency**: AI agents can request precisely the data needed in a single GraphQL query, reducing API calls, network overhead, and token usage. This focused approach delivers faster responses and lower operational costs compared to orchestrating multiple REST endpoints.

Check warning on line 106 in docs/source/how-it-works.mdx

View check run for this annotation

Apollo Librarian / AI Style Review

docs/source/how-it-works.mdx#L106 

Remove the emoji, as it is not in the approved set. Remove bolding, as it should not be used for emphasis.

```suggestion
Efficiency: AI agents can request precisely the data needed in a single GraphQL query, reducing API calls, network overhead, and token usage. This focused approach delivers faster responses and lower operational costs compared to orchestrating multiple REST endpoints.
```

**🔄 Agility**: The pace of AI development demands infrastructure that can evolve daily. GraphQL's declarative approach lets teams rapidly create, modify, and deploy new AI capabilities through self-service tooling. Product teams can wire up new MCP tools without waiting for custom development, keeping pace with AI's unprecedented velocity.

Check notice on line 108 in docs/source/how-it-works.mdx

View check run for this annotation

Apollo Librarian / AI Style Review

docs/source/how-it-works.mdx#L108 

Replace the idiomatic phrase 'wire up' with a simpler verb like 'add'.

```suggestion
**🔄 Agility**: The pace of AI development demands infrastructure that can evolve daily. GraphQL's declarative approach lets teams rapidly create, modify, and deploy new AI capabilities through self-service tooling. Product teams can add new MCP tools without waiting for custom development, keeping pace with AI's unprecedented velocity.
```

With Apollo MCP Server, these GraphQL advantages become immediately accessible to AI applications through standardized MCP tools.

## Benefits of Apollo MCP Server

Check warning on line 112 in docs/source/how-it-works.mdx

View check run for this annotation

Apollo Librarian / AI Style Review

docs/source/how-it-works.mdx#L112 

For conceptual documents, headings should use gerunds for consistency.

```suggestion
## Reviewing the benefits
```

- **🤖 Enable AI-enabled API orchestration**. With Apollo MCP Server, AI models can act as intelligent orchestrators of their GraphQL API operations. By exposing GraphQL operations as distinct MCP tools, AI clients can dynamically chain these operations together, in combination with other MCP servers and tools to execute complex workflows and automate multi-step processes.

Check notice on line 114 in docs/source/how-it-works.mdx

View check run for this annotation

Apollo Librarian / AI Style Review

docs/source/how-it-works.mdx#L114 

Remove the emoji, as it is not in the approved set. Use a colon after the bolded phrase to separate it from the description.

```suggestion
- **Enable AI-enabled API orchestration**: With Apollo MCP Server, AI models can act as intelligent orchestrators of their GraphQL API operations. By exposing GraphQL operations as distinct MCP tools, AI clients can dynamically chain these operations together, in combination with other MCP servers and tools to execute complex workflows and automate multi-step processes. 
```

- **🚀 Connect AI to GraphQL in Minutes**. Developers can expose existing or new GraphQL API operations to AI clients without building complex custom integrations. By translating GraphQL functionalities into standardized MCP tools, Apollo MCP Server can significantly reduce the effort needed to connect AI to diverse data sources.

Check warning on line 116 in docs/source/how-it-works.mdx

View check run for this annotation

Apollo Librarian / AI Style Review

docs/source/how-it-works.mdx#L116 

Avoid exaggerated claims like 'in Minutes'. Use sentence case for the bolded phrase and a colon to separate it from the description.

```suggestion
- **Connect AI to GraphQL quickly**: Developers can expose existing or new GraphQL API operations to AI clients without building complex custom integrations. By translating GraphQL functionalities into standardized MCP tools, Apollo MCP Server can significantly reduce the effort needed to connect AI to diverse data sources.
```

- **🔒 Maintain Full Security Control**. By using pre-defined, pre-approved persisted queries, developers can maintain precise governance over which data and operations AI clients can access. This ensures that AI uses existing security protocols and data access policies.

Check notice on line 118 in docs/source/how-it-works.mdx

View check run for this annotation

Apollo Librarian / AI Style Review

docs/source/how-it-works.mdx#L118 

Use sentence case for the bolded phrase and a colon to separate it from the description. Frame content relative to the reader by using 'you'.

```suggestion
- **Maintain full security control**: By using pre-defined, pre-approved persisted queries, you can maintain precise governance over which data and operations AI clients can access. This ensures that AI uses existing security protocols and data access policies.
```

## Prerequisites

- A GraphQL API
- An MCP Client

## Getting started

Ready to connect AI to your GraphQL API? Follow our [5-minute quickstart](/apollo-mcp-server/quickstart) to see Apollo MCP Server in action, or explore the [config file reference](/apollo-mcp-server/config-file) for detailed configuration options.

Check warning on line 127 in docs/source/how-it-works.mdx

View check run for this annotation

Apollo Librarian / AI Style Review

docs/source/how-it-works.mdx#L127 

Avoid making exaggerated time-based claims like '5-minute quickstart', as this can be frustrating for readers if it takes them longer.

```suggestion
Ready to connect AI to your GraphQL API? Follow our [quickstart](/apollo-mcp-server/quickstart) to see Apollo MCP Server in action, or explore the [config file reference](/apollo-mcp-server/config-file) for detailed configuration options.
```
220 changes: 98 additions & 122 deletions docs/source/index.mdx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -12,125 +12,101 @@

</PreviewFeature>

Apollo MCP Server provides a standard way for AI models to access and orchestrate your APIs running with Apollo.

## What is MCP?

[Model Context Protocol (MCP)](https://modelcontextprotocol.io/introduction) is an open protocol that standardizes how applications provide context to AI models like Large Language Models (LLM). MCP enables LLMs and AI agents to indirectly fetch data from external sources.

MCP follows a client-server architecture. MCP servers expose functions, called _tools_, that MCP clients can invoke.

## What is Apollo MCP Server?

Apollo MCP Server is an implementation of an MCP server. It makes GraphQL API operations available to AI clients as MCP tools. You can use Apollo MCP Server with any GraphQL API.

The GraphQL operations can be configured from persisted queries, which are predefined, approved lists of operations that are registered with and maintained by a graph. The operations can also be determined by AI introspecting your graph schema.

Apollo MCP Server is deployable in local environments via Apollo's Rover CLI or in containerized services in your cloud infrastructure. It can expose an MCP endpoint using Streamable HTTP for communication with AI clients.

## How Apollo MCP Server works

Apollo MCP Server bridges AI applications and your GraphQL APIs, translating GraphQL operations into MCP tools that AI models can discover and use.

```mermaid
graph LR
%% Nodes
AI["AI Application\n(Claude, ChatGPT, etc.)"]
MCPClient["MCP Client\n(Built into AI app)"]
MCPServer["Apollo MCP Server"]
GraphQL["GraphQL API\n(Your Graph)"]
Data["Your Data Sources\n(Databases, APIs, etc.)"]

%% Connections
AI <-->|"Natural Language\nRequests"| MCPClient
MCPClient <-->|"MCP Protocol\n(stdio/Streamable HTTP)"| MCPServer
MCPServer <-->|"GraphQL\nOperations"| GraphQL
GraphQL <-->|"Data\nQueries"| Data

%% Tool Generation
subgraph ToolGeneration[Tool Generation]
direction TB
OpFiles["Operation Files\n(.graphql)"]
PQM["Persisted Query\nManifests"]
Introspection["Schema\nIntrospection"]
Tools["MCP Tools"]

OpFiles --> Tools
PQM --> Tools
Introspection --> Tools
end

MCPServer -.->|"Exposes"| Tools
Tools -.->|"Available to"| MCPClient

%% Styling
classDef default stroke-width:1px
classDef aiClient stroke-width:2px
classDef mcpComponent stroke-width:2px
classDef apolloComponent stroke-width:2px
classDef apiComponent stroke-width:2px
classDef dataComponent stroke-width:2px

class AI aiClient
class MCPClient mcpComponent
class MCPServer apolloComponent
class GraphQL apiComponent
class Data dataComponent
class OpFiles,PQM,Introspection apolloComponent
class Tools mcpComponent
```

The architecture enables intelligent API orchestration through these components:

* AI Applications: Tools like Claude Desktop or ChatGPT connect to Apollo MCP Server through their built-in MCP clients, making requests in natural language.
* Transport Options: Communication happens over stdio for local development or Streamable HTTP.
* Tool Generation: Apollo MCP Server creates MCP tools from your GraphQL operations using:
* Operation Files: Individual `.graphql` files for specific queries or mutations
* Persisted Query Manifests: Pre-approved operation lists from Apollo GraphOS
* Schema Introspection: Dynamic operation discovery for flexible AI exploration

* Secure Execution: When invoked, the server executes GraphQL operations against your API endpoint, respecting all existing authentication, headers, and security policies.
* Existing Infrastructure: Your GraphQL API handles requests normally, with Apollo MCP Server acting as a controlled gateway rather than requiring any changes to your graph.

This design lets you expose precise GraphQL capabilities to AI while maintaining complete control over data access and security.

### Example usage

Once configured, AI applications can use your GraphQL operations naturally:

> User: "Show me the astronauts currently in space"
>
> Claude: *Uses GetAstronautsCurrentlyInSpace tool to query your GraphQL API*
>
> "There are currently 7 astronauts aboard the ISS..."

## Why GraphQL for AI?

GraphQL's architecture provides unique advantages for AI-powered API orchestration:

**🎯 Deterministic Execution**: GraphQL's built-in relationship handling and query structure eliminate guesswork for AI models. The graph defines clear paths between data types, ensuring AI agents execute operations in the correct sequence without complex prompt engineering or error-prone orchestration logic.

**🛡️ Policy Enforcement**: Security policies and access controls apply consistently across all services within a single GraphQL query context. This unified enforcement model ensures AI operations respect organizational boundaries, even when spanning multiple underlying APIs or microservices.

**⚡ Efficiency**: AI agents can request precisely the data needed in a single GraphQL query, reducing API calls, network overhead, and token usage. This focused approach delivers faster responses and lower operational costs compared to orchestrating multiple REST endpoints.

**🔄 Agility**: The pace of AI development demands infrastructure that can evolve daily. GraphQL's declarative approach lets teams rapidly create, modify, and deploy new AI capabilities through self-service tooling. Product teams can wire up new MCP tools without waiting for custom development, keeping pace with AI's unprecedented velocity.

With Apollo MCP Server, these GraphQL advantages become immediately accessible to AI applications through standardized MCP tools.

## Benefits of Apollo MCP Server

- **🤖 Enable AI-enabled API orchestration**. With Apollo MCP Server, AI models can act as intelligent orchestrators of their GraphQL API operations. By exposing GraphQL operations as distinct MCP tools, AI clients can dynamically chain these operations together, in combination with other MCP servers and tools to execute complex workflows and automate multi-step processes.

- **🚀 Connect AI to GraphQL in Minutes**. Developers can expose existing or new GraphQL API operations to AI clients without building complex custom integrations. By translating GraphQL functionalities into standardized MCP tools, Apollo MCP Server can significantly reduce the effort needed to connect AI to diverse data sources.

- **🔒 Maintain Full Security Control**. By using pre-defined, pre-approved persisted queries, developers can maintain precise governance over which data and operations AI clients can access. This ensures that AI uses existing security protocols and data access policies.

## Prerequisites

- A GraphQL API
- An MCP Client

## Getting started

Ready to connect AI to your GraphQL API? Follow our [5-minute quickstart](/apollo-mcp-server/quickstart) to see Apollo MCP Server in action, or explore the [config file reference](/apollo-mcp-server/config-file) for detailed configuration options.
Apollo MCP Server provides a standard way for AI models to access and orchestrate your Graph APIs - by making GraphQL API operations available to AI clients as MCP tools. You can use Apollo MCP Server with any GraphQL API.

Check notice on line 15 in docs/source/index.mdx

View check run for this annotation

Apollo Librarian / AI Style Review

docs/source/index.mdx#L15 

Use 'GraphQL APIs' for consistency. Rephrase to avoid the em-dash for better readability.

```suggestion
Apollo MCP Server provides a standard way for AI models to access and orchestrate your GraphQL APIs. It makes GraphQL API operations available to AI clients as MCP tools. You can use Apollo MCP Server with any GraphQL API. 
```

<!-- Benefits -->
- **AI-enabled API orchestration**: AI models can act as intelligent orchestrators of GraphQL API operations, dynamically chaining operations to execute complex workflows and automate multi-step processes.
- **More Efficient Tools for Agents**:Combining MCP tools and GraphQL operations means smarter schema search and discovery - reducing context usage and improving agent efficiency. [Learn more](https://www.apollographql.com/blog/smart-schema-discovery-how-apollo-mcp-server-maximizes-ai-context-efficiency)

Check failure on line 19 in docs/source/index.mdx

View check run for this annotation

Apollo Librarian / AI Style Review

docs/source/index.mdx#L19 

Correct the typo in 'Combining'. Use sentence case for the bolded phrase for consistency. Replace the vague link text 'Learn more' with a more descriptive phrase.

```suggestion
- **More efficient tools for agents**: Combining MCP tools and GraphQL operations enables smarter schema search and discovery, which reduces context usage and improves agent efficiency. [Read our blog post on maximizing AI context efficiency](https://www.apollographql.com/blog/smart-schema-discovery-how-apollo-mcp-server-maximizes-ai-context-efficiency)
```

- **Maintain Full Security Control**: Use pre-defined, pre-approved persisted queries to maintain precise governance over which data and operations AI clients can access.

Check notice on line 21 in docs/source/index.mdx

View check run for this annotation

Apollo Librarian / AI Style Review

docs/source/index.mdx#L21 

Use sentence case for the bolded phrase for consistency with other list items.

```suggestion
- **Maintain full security control**: Use pre-defined, pre-approved persisted queries to maintain precise governance over which data and operations AI clients can access.
```


<!-- Getting Started CTA -->
<div class="mb-16 rounded-lg border border-blue-200 bg-gradient-to-r from-blue-50 to-indigo-50 p-8">
<h2 class="mb-4 text-2xl font-bold">Getting Started</h2>
<p class="mb-6 text-gray-700">Get up and running in a couple of minutes with the Quickstart. Topics include how to:

Check warning on line 27 in docs/source/index.mdx

View check run for this annotation

Apollo Librarian / AI Style Review

docs/source/index.mdx#L27 

Avoid vague and exaggerated timeframes like 'in a couple of minutes'. The word 'couple' is also discouraged.

```suggestion
  <p class="mb-6 text-gray-700">Get up and running quickly with the Quickstart. It covers how to:
```

<ul class="list-inside list-disc space-y-2 text-gray-700 mb-6">
<li>creating your own MCP server (starting from scratch or using an existing Graph)</li>

Check notice on line 30 in docs/source/index.mdx

View check run for this annotation

Apollo Librarian / AI Style Review

docs/source/index.mdx#L30 

The word 'graph' should not be capitalized unless it is part of a proper noun.

```suggestion
    <li>creating your own MCP server (starting from scratch or using an existing graph)</li>
```
<li>connect and test your MCP server in an AI client application (like Claude or Cursor)</li>

Check notice on line 31 in docs/source/index.mdx

View check run for this annotation

Apollo Librarian / AI Style Review

docs/source/index.mdx#L31 

List items should use a consistent structure. Use gerunds ('-ing' verbs) to match the other items in this list.

```suggestion
    <li>connecting and testing your MCP server in an AI client application (like Claude or Cursor)</li>
```
<li>exposing your GraphQL API operations as MCP tools </li>
</ul>

</p>
<div class="flex gap-4">
<a href="/docs/apollo-mcp-server/quickstart" class="inline-block rounded-lg bg-blue-600 px-6 py-3 font-semibold text-white transition hover:bg-blue-700">Quickstart → </a>
</div>
</div>

<!-- Feature Cards -->
<div class="mb-12">
<h2 class="mb-6 text-2xl font-bold">Key Features & Guides</h2>

<div class="grid w-full grid-cols-2 gap-4">

<div class="rounded-lg border border-gray-200 p-5 transition hover:shadow-md">
<div class="text-3xl">📘</div>
<div>
<h3 class="mb-2 font-semibold">
<a href="/docs/apollo-mcp-server/how-it-works" class="text-blue-600 hover:underline">How It Works</a>

Check notice on line 51 in docs/source/index.mdx

View check run for this annotation

Apollo Librarian / AI Style Review

docs/source/index.mdx#L51 

The link text should match the casing of the destination page's title, which is 'How it Works'.

```suggestion
            <a href="/docs/apollo-mcp-server/how-it-works" class="text-blue-600 hover:underline">How it Works</a>
```
</h3>
<p class="text-sm text-gray-600">Understand the MCP architecture and how Apollo MCP Server works.</p>
</div>
</div>

<!-- Running MCP Server -->
<div class="rounded-lg border border-gray-200 p-5 transition hover:shadow-md">
<div class="mb-3 text-2xl">🚀</div>
<h3 class="mb-2 font-semibold">
<a href="/docs/apollo-mcp-server/running" class="text-blue-600 hover:underline">Running the MCP Server</a>
</h3>
<p class="text-sm text-gray-600">Run the server locally with <code>rover dev</code> or in containerized services.</p>
</div>

<!-- Tool Definition -->
<div class="rounded-lg border border-gray-200 p-5 transition hover:shadow-md">
<div class="mb-3 text-2xl">🔧</div>
<h3 class="mb-2 font-semibold">
<a href="/docs/apollo-mcp-server/tool-definition" class="text-blue-600 hover:underline">Tool Definition</a>
</h3>
<p class="text-sm text-gray-600">Create MCP tools from operation files, persisted queries, or schema introspection.</p>
</div>

<!-- Configuration -->
<div class="rounded-lg border border-gray-200 p-5 transition hover:shadow-md">
<div class="mb-3 text-2xl">⚙️</div>
<h3 class="mb-2 font-semibold">
<a href="/docs/apollo-mcp-server/configuration" class="text-blue-600 hover:underline">Configuration</a>
</h3>
<p class="text-sm text-gray-600">Configure transport options, endpoints, and GraphQL operations.</p>
</div>

<!-- Deploying -->
<div class="rounded-lg border border-gray-200 p-5 transition hover:shadow-md">
<div class="mb-3 text-2xl">☁️</div>
<h3 class="mb-2 font-semibold">
<a href="/docs/apollo-mcp-server/deploying" class="text-blue-600 hover:underline">Deployment</a>
</h3>
<p class="text-sm text-gray-600">Deploy to cloud infrastructure with containerization guides.</p>
</div>

<!-- Telemetry -->
<div class="rounded-lg border border-gray-200 p-5 transition hover:shadow-md">
<div class="mb-3 text-2xl">📊</div>
<h3 class="mb-2 font-semibold">
<a href="/docs/apollo-mcp-server/telemetry" class="text-blue-600 hover:underline">Telemetry</a>
</h3>
<p class="text-sm text-gray-600">Observe your MCP server requests and performance.</p>
</div>

<!-- Authentication -->
<div class="rounded-lg border border-gray-200 p-5 transition hover:shadow-md">
<div class="mb-3 text-2xl">🔐</div>
<h3 class="mb-2 font-semibold">
<a href="/docs/apollo-mcp-server/authentication" class="text-blue-600 hover:underline">Auth</a>

Check notice on line 106 in docs/source/index.mdx

View check run for this annotation

Apollo Librarian / AI Style Review

docs/source/index.mdx#L106 

Use the full word 'Authentication' for the link text and page title for clarity, instead of the abbreviation 'Auth'.

```suggestion
          <a href="/docs/apollo-mcp-server/authentication" class="text-blue-600 hover:underline">Authentication</a>
```
</h3>
<p class="text-sm text-gray-600">Secure your MCP server with headers and authorization.</p>
</div>
</div>
</div>
</div>