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

# Connecting to the MCP Server

> Add the hosted Credit Benchmark MCP server to Claude, Microsoft 365 Copilot, ChatGPT, Claude Code, Cursor, VS Code, or any MCP-compatible client using OAuth sign-in.

> [!NOTE]
>
> Your Credit Benchmark account must be enabled for MCP access before you connect. Access is not enabled automatically. See [Request access](/mcp/overview#request-access).

Every client connects to the same server URL:

```text theme={null}
https://mcp.creditbenchmark.com/mcp
```

Every client authenticates with **OAuth**. There is no API key or client secret to copy:

1. Your client registers itself with the server automatically.
2. You are redirected to the Credit Benchmark sign-in page (Okta) and sign in with your Credit Benchmark account.
3. Your client receives an access token and refreshes it automatically.

If a client asks for an authentication type, choose **OAuth**. Leave any OAuth client ID and client secret fields empty. The only exception is the Microsoft 365 Copilot federated connector, which uses a client ID and secret from Credit Benchmark.

**Claude**

Works in Claude on the web and in Claude Desktop.

1. Open **Settings → Connectors** and select **Add custom connector**.
2. Enter a name, such as `Credit Benchmark`, and the URL `https://mcp.creditbenchmark.com/mcp`. Leave the OAuth client ID and secret under **Advanced settings** empty.
3. Select **Add**, then **Connect**.
4. Sign in with your Credit Benchmark account on the OAuth sign-in page that opens.
5. Check that the connector is switched on.

> [!NOTE]
>
> On Team and Enterprise plans, an organisation owner may need to add the connector first. Each member then connects and signs in with their own Credit Benchmark account.

**Microsoft 365 Copilot**

There are two ways to use Credit Benchmark in Microsoft 365 Copilot:

| | Federated connector (recommended) | Copilot Studio agent |
| - | - | - |
| **Where it works** | Copilot Chat, agents, Researcher, and Copilot in Word, Excel, PowerPoint, and Outlook | Inside the agent you build, in Copilot and Teams |
| **Who sets it up** | A Microsoft 365 admin, once for the tenant | A Copilot Studio maker, once per agent |
| **Model** | Copilot's default model | You can select a Claude model for the agent |

In both options, each user signs in with their own Credit Benchmark account through OAuth, so results follow their own entitlements.

### Option 1: Federated connector (recommended)

A federated connector makes Credit Benchmark available as a data source across Microsoft 365 Copilot. Data is fetched live through MCP and isn't indexed into Microsoft 365.

**Admin setup** (Global Administrator or AI Administrator):

1. Request the OAuth **Client ID** and **Client secret** for Microsoft 365 Copilot from [support@creditbenchmark.com](mailto:support@creditbenchmark.com).
2. In the [Teams Developer Portal](https://dev.teams.microsoft.com/), select **Tools → OAuth Client Registration → New OAuth connection**, and enter:

   | Field | Value |
   | - | - |
   | **Name** | `Credit Benchmark` |
   | **Client ID** | The client ID from Credit Benchmark |
   | **Client secret** | The client secret from Credit Benchmark |
   | **Authorization endpoint** | `https://mcp.creditbenchmark.com/authorize` |
   | **Token endpoint** | `https://mcp.creditbenchmark.com/token` |
   | **Refresh endpoint** | `https://mcp.creditbenchmark.com/token` |
   | **Scopes** | `openid profile email offline_access` |

   Turn on **Proof Key for Code Exchange (PKCE)**, select **Save**, and copy the **OAuth client registration ID**.
3. In the [Microsoft 365 admin center](https://admin.microsoft.com/), go to **Copilot → Connectors**. On the **Gallery** tab, under **Created by your org**, select **Add** on the **Create a new connector** tile.
4. On the **Custom connector** page, under **Connect to MCP server**, select **Add**. Enter:
   * **Display name:** `Credit Benchmark`
   * **Base URL:** `https://mcp.creditbenchmark.com/mcp`
   * **OAuth registration ID:** the ID you copied in step 2
5. Select **Save**. Use **Staged rollout** to release the connector to a test group first, then select **Deploy to all users**. Changes can take up to 15 minutes to apply.

**Each user:**

1. In Microsoft 365 Copilot Chat, select the **...** menu, then **Settings → Sources**.
2. Find **Credit Benchmark**, select **Connect**, and sign in with your Credit Benchmark account on the OAuth sign-in page.
3. Ask your question in Copilot Chat, or select Credit Benchmark from **Sources** in the Researcher agent.

See Microsoft's [federated connector setup guide](https://learn.microsoft.com/en-us/microsoft-365/copilot/connectors/set-up-custom-federated-connectors) for more detail.

### Option 2: Copilot Studio agent

Use this option to build a dedicated Credit Benchmark agent, or to run it on a Claude model.

**Agent setup** (a Copilot Studio maker or admin):

1. In [Copilot Studio](https://copilotstudio.microsoft.com), create an agent or open an existing one.
2. On the agent's **Tools** page, select **Add a tool → New tool → Model Context Protocol**.
3. Enter a name, such as `Credit Benchmark`, a short description, and the server URL `https://mcp.creditbenchmark.com/mcp`.
4. Set **Authentication** to **OAuth 2.0** and choose **Dynamic discovery**, then select **Create**. Leave any client ID and client secret fields empty.
5. Create a connection when prompted and sign in with a Credit Benchmark account on the OAuth sign-in page, then add the tool to the agent.
6. In the tool's settings, set **Credentials to use** to **End user credentials**, so each person signs in with their own Credit Benchmark account.
7. Optional: under the agent's model settings, select a Claude model.
8. Publish the agent to the **Microsoft 365 Copilot** and **Teams** channels.

**Each user:**

1. Open the agent from the agents list in Microsoft 365 Copilot or Teams.
2. The first time it calls a Credit Benchmark tool, select **Sign in** and sign in with your own Credit Benchmark account on the OAuth sign-in page.

> [!NOTE]
>
> Federated connectors, available models, and agent publishing depend on your Microsoft 365 tenant settings. Ask your Microsoft 365 administrator if you cannot add connectors or tools.

**ChatGPT**

1. Open **Settings → Apps & Connectors**.
2. Under **Advanced settings**, turn on **Developer mode**.
3. Select **Create**, enter a name, such as `Credit Benchmark`, and the MCP server URL `https://mcp.creditbenchmark.com/mcp`.
4. Set **Authentication** to **OAuth**, then select **Create**.
5. Sign in with your Credit Benchmark account on the OAuth sign-in page that opens.
6. In a new chat, enable the connector from the tools menu.

> [!NOTE]
>
> Custom connectors depend on your ChatGPT plan and workspace settings. If you cannot see the option, ask your workspace administrator.

**Claude Code**

1. Add the server from your terminal:

   ```bash theme={null}
   claude mcp add --transport http credit-benchmark https://mcp.creditbenchmark.com/mcp
   ```

   Add `--scope user` to make it available in every project, or `--scope project` to share it with your team through the project's `.mcp.json`.

2. Start Claude Code and run `/mcp`.

3. Select `credit-benchmark`, choose **Authenticate**, and sign in with your Credit Benchmark account on the OAuth sign-in page that opens in your browser.

**Cursor**

1. Add the server to `~/.cursor/mcp.json` for all projects, or to `.cursor/mcp.json` for one project:

   ```json theme={null}
   {
     "mcpServers": {
       "credit-benchmark": {
         "url": "https://mcp.creditbenchmark.com/mcp"
       }
     }
   }
   ```

   No headers or API keys are needed. Cursor uses OAuth.

2. Open **Cursor Settings → MCP**. `credit-benchmark` shows that it needs a login.

3. Select **Login** and sign in with your Credit Benchmark account on the OAuth sign-in page that opens in your browser.

**VS Code**

1. Add the server to `.vscode/mcp.json` in your workspace, or run **MCP: Add Server** from the Command Palette and choose **HTTP**:

   ```json theme={null}
   {
     "servers": {
       "credit-benchmark": {
         "type": "http",
         "url": "https://mcp.creditbenchmark.com/mcp"
       }
     }
   }
   ```

   No headers or API keys are needed. VS Code uses OAuth.

2. Start the server from the `mcp.json` editor or the **Extensions** view.

3. When VS Code asks to authenticate, allow it and sign in with your Credit Benchmark account on the OAuth sign-in page that opens in your browser.

The tools are then available in Copilot Chat agent mode.

**Other clients**

Any MCP client that supports the Streamable HTTP transport and OAuth 2.1 can connect:

1. Point the client at `https://mcp.creditbenchmark.com/mcp`.
2. The client discovers the OAuth endpoints from the server's discovery metadata and registers itself through dynamic client registration ([RFC 7591](https://datatracker.ietf.org/doc/html/rfc7591)).
3. The client sends you through the authorization code flow with PKCE, where you sign in with your Credit Benchmark account.
4. The client sends the resulting access token as a bearer token in the `Authorization` header of every request, and refreshes it when it expires.

## Check the connection

In a new chat, ask:

> What's the credit picture for Tesco?

Your assistant should show that it is calling a Credit Benchmark tool, then reply with a short credit summary based on the current consensus.

## Troubleshooting

| Symptom | What to do |
| - | - |
| Sign-in fails, or every tool call returns an access error | Your account may not be enabled for MCP access yet. See [Request access](/mcp/overview#request-access). |
| The assistant does not use Credit Benchmark | Check that the connector is added and switched on for the chat. If it shows an error, select it for details. |
| A tool result has `category: auth_failed` or `auth_missing_credentials` | Sign-in did not complete. Disconnect the connector, connect again, and sign in. |
| A tool result has `category: upstream_unavailable` | The server could not reach Credit Benchmark. Check your network or VPN and retry. |
| Tool calls hang | A token refresh is probably stuck. Remove the connector, add it again, and sign in. |
| Results are empty for a name you expect | The entity may not have consensus coverage, or may sit outside your entitlements. Try the Web App, or ask about the entity by LEI. |

If a problem continues, email [support@creditbenchmark.com](mailto:support@creditbenchmark.com) with the approximate time of the failed request.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.