# Best Way to Connect a Remote MCP Server to Claude in 2026
> Connect a remote MCP server to Claude, complete OAuth authorization, enable its tools in chat, test the connection, and fix common connector setup errors.
- **Author**: parth-kanpariya
- **Published**: 2026-09-24
- **Modified**: 2026-09-24
- **Category**: AI & DevOps
- **URL**: https://kuberns.com/blogs/connect-remote-mcp-server-to-claude/
---
To connect a remote MCP server to Claude, open **Settings → Connectors**, choose **Add custom connector**, enter the server's public HTTPS endpoint and complete the required authentication. After the connector is authorized, enable it in a Claude conversation and run a safe request to verify that Claude can access its tools.
For Kuberns, use `https://mcp.kuberns.cloud/mcp` as the remote endpoint. The connection lets Claude use the Kuberns tools available to the authorized account to inspect deployment status, review logs, examine non-secret configuration and troubleshoot applications without manually copying that context into a conversation.
This guide covers the regular Claude web and desktop connector experience. Developers working inside Claude Code should use the separate Claude Code MCP setup guide once it is available. If you built a server that still needs a public endpoint, follow the guide to [deploy your MCP server to production](https://kuberns.com/blogs/deploy-mcp-server/) before connecting it here.
## TL;DR
- Open Claude **Settings → Connectors**.
- Select **+ → Add custom connector**.
- Enter `https://mcp.kuberns.cloud/mcp` for Kuberns.
- Keep the detected OAuth authentication settings and add the connector.
- Select **Connect**, sign in to Kuberns and review the permissions.
- Enable Kuberns from the **+ → Connectors** menu in a Claude conversation.
- Ask Claude to use Kuberns for a safe, read-only deployment-status request.
## What Can Claude Do With the Kuberns MCP Server?
Claude provides the conversational interface, while the Kuberns MCP server supplies the deployment data and operational tools authorized for the connected account. The exact capabilities come from the current tool inventory, not from Claude itself.
The permissions displayed during the tested Kuberns connection support these workflows:
| Deployment task | Example Claude request |
|---|---|
| Check deployment status | “Use Kuberns to check the status of my latest deployment.” |
| Review build history | “Summarize the last three builds for my application.” |
| Diagnose a failed build | “Inspect my latest failed build and identify the first actionable error.” |
| Review runtime logs | “Find recurring errors in the recent runtime logs for my backend.” |
| Check configuration safely | “List the configured environment variable names without displaying their values.” |
| Review resource configuration | “Show the current resource configuration for this application.” |
| Check billing information | “Show my current Kuberns credits and billing information.” |
| Contact support | “Create a support ticket using the relevant details from this failed build.” |
These examples do not mean every MCP server offers the same tools. Claude can use only the capabilities published by the selected server and approved by the user. Start with a read-only request, verify the returned account and project, and review any confirmation before allowing an action that writes or changes data.
## What Do You Need Before Connecting Claude to Kuberns?
Before starting, confirm that you have:
- A Claude account with access to custom connectors.
- Permission to add a connector in the selected Claude workspace.
- A Kuberns account that you can authorize.
- At least one Kuberns deployment for a meaningful test.
- The public Kuberns MCP endpoint: `https://mcp.kuberns.cloud/mcp`.
Anthropic's current custom connector documentation says remote MCP connectors are available through supported Claude experiences, but administration differs by account type. Individual users can add a connector from their own Connectors settings. Team and Enterprise organizations may require an Owner to add the connector before members can authenticate it.
A remote connector must also be reachable from Anthropic's cloud infrastructure. Being able to open an endpoint from your own computer does not prove that Claude can reach it. Anthropic explains that account-based remote connectors originate from its cloud even when the user is working in Claude Desktop.
> **Nothing deployed on Kuberns yet?** [Deploy an application from your GitHub repository](https://dashboard.kuberns.com/) first, then connect Claude to its supported deployment tools.
## How to Connect a Remote MCP Server to Claude
The setup uses Claude's account-level Connectors interface. You do not need to edit a local Claude Desktop configuration file for this remote Kuberns connection.
### Step 1: Open Claude Connectors
Open Claude Settings and select **Connectors**. Click the **+** button, then choose **Add custom connector**.
On Team or Enterprise accounts, an Owner may need to add the remote connector through the organization settings first. Once it is available to the organization, individual members can connect their own authorized accounts.
**Expected result:** Claude opens a form that accepts a connector name and a remote MCP URL.
### Step 2: Enter the Kuberns Remote MCP Endpoint
Enter the following details:
| Field | Value |
|---|---|
| Name | Kuberns |
| Remote MCP URL | `https://mcp.kuberns.cloud/mcp` |
Confirm that the URL uses HTTPS, the hostname is `mcp.kuberns.cloud` and the path ends in `/mcp`. Only connect an endpoint published by a provider you trust.
Select **Continue** after verifying the URL.
**Expected result:** Claude contacts the endpoint and displays the authentication configuration it detected.
### Step 3: Review the Detected Authentication Settings
The Kuberns connector uses an account authorization flow. In the interface used for this guide, Claude detected:
- **Sign in now** for authentication.
- **Register automatically** for the OAuth client.
These settings allow Claude to begin the authorization process after the connector is created. Do not add request headers, API keys or manually created OAuth credentials unless current Kuberns documentation explicitly requires them.
Review the detected settings, then add the connector.
**Expected result:** Kuberns appears in the Claude Connectors list with an option to connect the account.
### Step 4: Connect Your Kuberns Account
Open the newly added Kuberns connector and select **Connect**. Claude redirects you to the Kuberns authorization page so that you can sign in and decide whether to grant the requested access.
This redirect is expected. OAuth lets the user authorize access without giving their Kuberns password directly to Claude. It does not remove the need to verify the requesting application, domain and permissions.
**Expected result:** The Kuberns consent page lists the specific access requested by the Claude connection.
### Step 5: Review and Approve the Permissions
The Kuberns authorization screen used for this guide requests permission to:
- View deployment status and build history.
- View build and runtime logs.
- View environment and resource configuration without secret values.
- View billing and credits information.
- Create and update support tickets.
- Debug the application.
Read the current screen before selecting **Approve**. Permissions and tool inventories can change, so the live consent page is the authoritative description of the access being requested.
The displayed configuration permission excludes secret values. Even so, avoid putting credentials, customer information or other unnecessary sensitive data into Claude prompts.
**Expected result:** Claude returns to the connector and shows Kuberns as connected.
### Step 6: Enable Kuberns in a Claude Conversation
Adding and authorizing a connector does not necessarily enable it for every conversation. Open a new Claude conversation, select the **+** button and open **Connectors**. Enable Kuberns for the conversation where you want Claude to use its tools.
Anthropic's documentation explains that connectors can be enabled or disabled per conversation. If Claude answers from general knowledge instead of checking the deployment, verify that Kuberns is enabled before changing the prompt.
**Expected result:** Kuberns appears among the tools available to the active conversation.
### Step 7: Run a Safe Test Request
Start with a read-only request that is easy to verify:
> Use Kuberns to show the deployment status and build history of my most recently deployed application.
A successful result should confirm that:
- Claude recognized the Kuberns connector.
- Account authorization completed successfully.
- Claude selected an available Kuberns tool.
- The returned information belongs to the connected Kuberns account.
- No secret values were displayed.
If the account contains several applications, include the application or service name in the prompt. Explicitly say “Use Kuberns” when Claude responds generally instead of invoking the connector.
## How Can You Use Claude for Deployment Troubleshooting?
Once the read-only test succeeds, use prompts that give Claude a specific application, timeframe and expected outcome.
**Check a deployment:**
- “Use Kuberns to check whether my latest production deployment completed successfully.”
- “Summarize the last three builds for my API service and identify any failure.”
**Investigate a failed build:**
- “Inspect the latest failed build and find the first meaningful error.”
- “Does this failure appear to come from dependencies, configuration or application code?”
If the logs reveal a familiar failure pattern, use the guide to [understand why software deployments fail](https://kuberns.com/blogs/why-do-software-deployments-fail/) before changing the application.
**Review runtime problems:**
- “Review the recent runtime logs and identify repeated errors.”
- “Look for database connection failures or missing configuration in my backend logs.”
**Review configuration without exposing secrets:**
- “List the environment variable names configured for this service without showing their values.”
- “Show the resource configuration for each service in this application.”
For the underlying deployment workflow, follow the recommended practices for [managing production environment variables safely](https://kuberns.com/blogs/environment-variables-in-production/).
**Prepare a support request:**
- “Summarize the relevant build failure and create a support ticket with the available context.”
Review the proposed action and its inputs before approving a tool that creates or changes external data.
## Can Claude Deploy or Modify an Application Through MCP?
Claude can perform only the operations exposed by the remote MCP server and permitted by the connected account. Adding a connector does not automatically give Claude the ability to deploy, restart, scale, roll back or modify an application.
The Kuberns flow verified for this guide exposes deployment status, build history, build and runtime logs, non-secret environment and resource configuration, billing and credits information, support-ticket operations and application debugging. Any additional action should be described only after the corresponding production tool is visible and tested.
Before approving a write operation, check the tool name, description, inputs and expected effect. After it runs, verify the resulting state in Kuberns rather than treating a conversational response as proof that the operation completed.
## Why Use the Kuberns Remote MCP Server With Claude?
Kuberns is a strong choice for developers who want Claude to work with real deployment context instead of a manually copied fragment. Claude provides the conversational interface, Kuberns supplies the authorized deployment information and supported operational tools, and MCP provides the standard connection between them.
This workflow reduces the time spent moving among a deployment dashboard, log viewer and AI conversation while investigating the same issue. Claude can request the supported context through the connector, subject to the user's permissions and tool approvals, while Kuberns remains the deployment platform behind the application.
[Kuberns](https://kuberns.com/) is an Agentic AI platform for deployment designed for full-stack and complex backend projects. New users can learn [how Kuberns deploys and manages applications](https://kuberns.com/blogs/what-is-kuberns-the-simplest-way-to-build-deploy-and-scale-full-stack-apps/) before connecting its remote MCP tools to Claude.
> Bring live deployment context into the Claude workflow you already use to investigate application problems. [Deploy your application on Kuberns](https://dashboard.kuberns.com/) and connect the remote MCP endpoint when it is ready.
## Is It Safe to Connect a Remote MCP Server to Claude?
A remote MCP connector can read information or perform supported operations using the access granted by the user. Treat it as an integration with an external system, not as a harmless interface preference.
Anthropic's security and privacy guidance for custom connectors recommends connecting only trusted servers, reviewing the requested permissions and paying attention to tool approvals. It also warns that a malicious or compromised connector could expose Claude to prompt injection or unexpected tool behavior.
Use these safeguards:
- Confirm the endpoint and provider before connecting.
- Review every permission on the authorization screen.
- Enable only the tools relevant to the current conversation.
- Examine tool inputs and outputs before approving an action.
- Avoid pasting secret values or unnecessary customer information into prompts.
- Verify consequential results in the source platform.
- Disconnect access when it is no longer required.
OAuth protects the sign-in flow, but it does not guarantee that every connected server or tool is safe. Trust still depends on who operates the server, what the tools do and what access the user grants.
## How to Fix Common Claude MCP Connection Problems
Use the point where the setup failed to narrow down the cause:
| Problem | Likely cause | What to do |
|---|---|---|
| Add custom connector is unavailable | Account, role or organization restriction | Check current Claude connector availability and ask the workspace Owner to add or enable it |
| Claude rejects the server URL | The URL is incorrect, private or unreachable from Anthropic's cloud | Confirm the public HTTPS endpoint and the provider's documented path |
| Authentication options are not detected | The endpoint or its authorization metadata cannot be read | Verify the MCP URL and recreate the connector |
| Kuberns appears but says it is not connected | The OAuth flow has not been completed | Open Kuberns in Connectors and select Connect |
| Authorization completes but tools do not appear | Authentication or tool discovery did not finish | Disconnect, reconnect and wait for the connector to load its tools |
| The connector is unavailable in a conversation | It is not enabled for that conversation | Select the plus menu, open Connectors and enable Kuberns |
| Claude answers without using Kuberns | The request is ambiguous or the connector is disabled | Explicitly ask Claude to use Kuberns and name the application |
| The connection repeatedly expires | Authorization was revoked or must be renewed | Reconnect Kuberns and complete the OAuth flow again |
| A local Desktop JSON setup does not work | Local MCP configuration is separate from account connectors | Add the remote server through Claude Connectors instead |
Community discussions demonstrate this confusion, but they should not replace the official instructions. One developer asked about connecting Claude Desktop to a remote MCP server after trying a local bridge. Another reported that OAuth completed but Claude did not establish the expected remote connection. These discussions validate the troubleshooting intent, while Anthropic's current documentation remains the technical authority.
## How to Disconnect the Remote MCP Server From Claude
Open **Settings → Connectors**, locate Kuberns and open its menu. Use the available option to disconnect or remove the connector. Anthropic notes that users can revoke authorization by disconnecting the connector in Claude or through the connected service's security settings.
Removing the connector stops it from being available in future Claude conversations. Disabling it for one conversation is different from revoking the account authorization, so use the full disconnect or remove control when access is no longer required.
If you reconnect later, review the live permissions again. Do not assume they are identical to the permissions granted during an earlier connection.
## Connect Claude to Your Kuberns Deployment
Connecting a remote MCP server to Claude requires a public HTTPS endpoint, an authorized account and a connector that is enabled for the active conversation. With the Kuberns endpoint connected, Claude can retrieve supported deployment status, logs, non-secret configuration and troubleshooting context without requiring the developer to assemble that information manually.
Deploy your application on Kuberns, connect `https://mcp.kuberns.cloud/mcp` through Claude Connectors and begin with a read-only deployment-status request. Expand to other tools only after reviewing their permissions and expected effects.
[](https://dashboard.kuberns.com/)
## Frequently Asked Questions
### How do I connect a remote MCP server to Claude?
Open Claude Settings, select Connectors, choose **Add custom connector** and enter the remote MCP server's public HTTPS URL. Add the connector, complete its authentication flow, enable it in a conversation and test one of its available tools.
### What URL should I use for the Kuberns MCP server?
Use `https://mcp.kuberns.cloud/mcp` as the remote MCP server URL when adding Kuberns as a custom connector in Claude.
### Can I use a remote MCP server in Claude Desktop?
Yes. Remote MCP connectors configured through Claude's Connectors settings are available through the supported Claude and Claude Desktop experience. This account-based connector flow is separate from local MCP servers configured on the computer.
### Why is my Claude connector added but not available in a conversation?
A configured connector may still need authentication and must be enabled for the conversation. Open the connector, complete its sign-in flow, then select **Connectors** from the plus menu in the conversation and enable it.
### Can Claude see Kuberns environment secret values?
The Kuberns authorization screen used for this guide permits Claude to view environment and resource configuration without secret values. Review the current consent screen before approving because it is the authoritative description of the requested access.
### Can Claude deploy or modify my Kuberns application?
Claude can perform only the operations exposed by the current Kuberns MCP tools and allowed by the connected account. Do not assume that the connector can deploy, restart, scale, roll back or modify an application unless the corresponding tool is visible and has been verified.
### How do I disconnect Kuberns from Claude?
Open Claude Settings, select Connectors, locate Kuberns and use its menu to disconnect or remove it. If an authorization-revocation control is available in Kuberns, revoke the connection there as well.
---
- [More AI & DevOps articles](https://kuberns.com/blogs/category/ai-devops/1/)
- [All articles](https://kuberns.com/blogs/)