> For the complete documentation index, see [llms.txt](https://help.nightfall.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://help.nightfall.ai/data-exfiltration-prevention/ai-agent-security/ai-governance/mcp-gateway/mcp-gateway-end-user-faqs.md).

# MCP Gateway End-User FAQs

## MCP Gateway End-User FAQs

This guide is written for employees - developers, engineers, data scientists, and other end-users - who have been invited to connect their AI tools to the **Nightfall MCP Gateway**.

***

### 1. Getting Started & Overview

#### What is Model Context Protocol (MCP)?

Model Context Protocol (MCP) is an open standard developed by Anthropic that allows AI assistants (such as Cursor, Claude, and VS Code) to securely interact with external tools, issue trackers, databases, and APIs. Instead of manually copying and pasting context into chat prompts, MCP allows your AI assistant to query data (like reading pull requests, inspecting Jira/Linear tickets, or searching internal documentation) on your behalf.

#### What is the Nightfall MCP Gateway?

The Nightfall MCP Gateway is a centralized, secure bridge between your AI clients and approved enterprise tools (such as GitHub, Linear, Notion, and internal APIs). Instead of having every developer generate personal API keys, configure separate local `.json` files, and manage individual tokens, the Gateway provides a **single setup URL** that securely routes and authenticates all approved tool calls through your organization's Single Sign-On (SSO).

#### Why did my Security / IT Team set this up for me?

Your security team provisioned this gateway to empower you with AI tools while eliminating friction and security risks:

* **No Plaintext API Tokens on Laptops:** Personal API keys and long-lived tokens no longer need to be generated or stored in local configuration files.
* **Streamlined Onboarding:** You get immediate access to all approved development tools with a single URL and one-click SSO login.
* **Safe Guardrails:** Destructive operations (such as deleting repositories or dropping database tables) are blocked at the gateway, enabling safe, high-speed read and query capabilities.

#### Who is Nightfall AI?

Nightfall AI is your organization's enterprise AI security and data security partner. Nightfall provides the zero-trust token brokering, granular permission scoping, and audit infrastructure powering the MCP Gateway.

***

### 2. Supported Clients & Environments

#### Which AI clients, IDEs, and tools are supported?

The MCP Gateway supports **any client, IDE, or agent framework** that supports standard Streamable HTTP / Server-Sent Events (SSE) and OAuth MCP connections, including:

* **Cursor IDE** (Composer, Chat & Agent)
* **Claude Desktop & Claude Code CLI**
* **VS Code** (via MCP extensions)
* **Windsurf IDE** (Cascade)
* **JetBrains IDEs** (via MCP plugins)
* **Claude.ai / ChatGPT Custom Connectors** (Enterprise / Team workspaces)
* **Custom AI Scripts & Autonomous Agents** (LangChain, LlamaIndex, AutoGen, custom Python/Node agents)

#### How long does setup take?

Setup takes less than **60 seconds**:

1. Click the activation link in your invite email and log in via your company SSO.
2. Copy your unique **Tenant MCP Endpoint URL** from the Setup tab.
3. Paste the URL into your AI client's MCP configuration settings.

***

### 3. Authentication, Tokens & Privacy

#### Do I need to create or manage personal API tokens for GitHub, Linear, etc.?

**No.** The gateway brokers credentials centrally. For OAuth-supported tools (like GitHub or Linear), you simply click **Connect** in the gateway once to authenticate via SSO. You never have to generate, copy, paste, or rotate personal access tokens.

#### Can administrators see my passwords or private credentials?

**No.** Stored tokens and OAuth secrets are encrypted at rest and injected directly into upstream tool calls. Administrators can see connection status (e.g., *Connected* or *Token Expired*), but they cannot view secrets, passwords, or personal keys.

#### Does the MCP Gateway or Nightfall read or train on my code and prompts?

**No.** Nightfall does not train AI models on your code, prompts, or tool data. The gateway functions strictly as an enterprise proxy. Tool calls and payloads are logged solely for your organization's security audit trail (with strict data retention limits).

***

### 4. Performance & Daily Workflow

#### Will the MCP Gateway slow down my AI assistant?

**No.** The gateway uses high-throughput, low-latency streaming HTTP connections. Tool discovery and invocations typically execute with sub-millisecond proxy overhead, ensuring your AI assistant responds instantaneously.

#### What happens if an upstream vendor (like GitHub or Linear) experiences an outage?

If a specific backend server encounters downtime, the gateway immediately surfaces the vendor's error message and short-circuits repeated failing calls. All other enabled tools and servers continue operating without disruption.

#### Can I customize or hide specific tools in my IDE?

**Yes.** On the **Enabled Servers** page in your Gateway dashboard, you can toggle **Enabled for me** on individual tools. If there are certain tools you prefer not to load into your assistant's context window, disabling them removes them exclusively from your client.

***

### 5. Troubleshooting & Support

#### Why does my AI client say "Tool not found" or fail to connect?

1. **Check SSO Authorization:** Ensure you have completed the one-time browser consent (`Approve`) when connecting your client to the gateway.
2. **Verify Upstream Connection:** If using a server like GitHub, verify that your account shows **Connected** under the Enabled Servers tab in the Gateway dashboard.
3. **Check Client URL:** Confirm that the endpoint URL in your client configuration matches the exact URL provided in your **Setup** tab.

#### Where can I get help or request access to new MCP servers?

* Reach out to your internal IT/Security team or post in your company's dedicated help channel
* For platform documentation and support, visit [help.nightfall.ai](https://help.nightfall.ai) or contact <support@nightfall.ai>.
