概覽

The Quicknode MCP server connects your Quicknode account to MCP-compatible AI assistants like ChatGPT, Codex, Claude Code, Claude Desktop, Cursor, VS Code, and more. Once connected, you can create and manage endpoints, monitor usage, configure security settings, and review billing data through natural language instead of navigating the dashboard or writing API scripts.

The server is hosted remotely at https://mcp.quicknode.com/mcp and authenticates via OAuth. It operates on top of Quicknode's 管理員 API, translating natural language instructions into API operations. The assistant interprets your request, selects the appropriate tool, and executes it after confirmation.

簡而言之

• Quicknode is an official connector in the Anthropic MCP directory — add it directly or find it under Customize > Connectors in Claude
• Quicknode is also available natively in ChatGPT 以及 《古籍》 — install directly from the ChatGPT Apps catalog or via /plugin in Codex
• All other clients (Cursor, VS Code, Windsurf, Zed) connect by pointing to https://mcp.quicknode.com/mcp and authorizing via OAuth
• For CI/CD and non-interactive environments, skip OAuth and pass an API key directly — see Using an API Key
• Manage endpoints, monitor usage, configure security rules, and review billing through natural language, no dashboard or API scripts required

Requirements

• A Quicknode account
• An MCP-compatible AI assistant (ChatGPT, Codex, Claude Code, Claude Desktop, Cursor, VS Code, Windsurf, Zed, or similar)

Configuration

Add the Quicknode MCP server to your assistant using the instructions for your client below.

克勞德

克勞德·韋布

Quicknode is an official connector in the Anthropic MCP directory. To connect it in Claude:

1. Visit the Quicknode connector page, or open Claude in your browser and navigate to Customize > Connectors to find Quicknode in the directory
2. Click Connect
3. A browser window will open — log in with your Quicknode account, select a role (Admin or Viewer), and approve access

Claude 桌面版

To add the Quicknode MCP server to the Claude Desktop app:

1. Navigate to Customize > Connectors
2. Find Quicknode in the directory and click Connect
3. A browser window will open — log in with your Quicknode account, select a role (Admin or Viewer), and approve access

克勞德・科德

To add the Quicknode MCP server to Claude Code, run the following command in your terminal:

1. Add the server:

claude mcp add --transport http quicknode https://mcp.quicknode.com/mcp

2. 啟動 Claude Code 並輸入 /mcp to authorize the server in the browser window that opens

Installation Scopes

By default, the command above installs the server in local scope (current project only). Claude Code supports three scopes:

• Local (default) — Available only to you in the current project
• Project (--scope project) — Shared with your team via a .mcp.json file in the project root
• User (--scope user) — Available to you across all projects

To install globally for your user, add the --scope user flag:

claude mcp add --transport http --scope user quicknode https://mcp.quicknode.com/mcp

OpenAIOpenAI

ChatGPT

To connect the Quicknode MCP server to ChatGPT:

1. Visit the Quicknode app on ChatGPT, or open ChatGPT, go to Apps in the sidebar, and search for Quicknode
2. Click Connect on the Quicknode app page
3. A browser window opens at mcp.quicknode.com where you log in with your Quicknode account, select a role (Admin 或 Viewer), and approve access
4. Quicknode is now available in your ChatGPT conversations. Mention @Quicknode in any chat to use it

《古籍》

To add Quicknode to Codex (OpenAI):

1. Open Codex and type /plugin in the input
2. Browse the available plugins and select Quicknode
3. Press Enter to view details, then select Install plugin
4. If you have not already connected via ChatGPT, a browser window opens for OAuth authorization where you log in with your Quicknode account, select a role (Admin or Viewer), and approve access
5. After installation, Quicknode tools are available in your Codex sessions

提示

If you have already connected Quicknode in ChatGPT, the authorization step is skipped and the plugin is ready to use immediately.

游標

To add the Quicknode MCP server to Cursor:

1. Go to Settings > Tools & MCPs, and click New MCP Server
2. (Optionally) Open (or create) .cursor/mcp.json in your project root or home directory
3. Add the following configuration:

{
  "mcpServers": {
    "quicknode": {
      "url": "https://mcp.quicknode.com/mcp"
    }
  }
}

4. Once you save the file, Cursor will detect the new server configuration.
5. Click Connect next to the Quicknode MCP server in the Cursor interface
6. A browser window will open where you can log in with your Quicknode account and approve access

VS Code

To add the Quicknode MCP server to VS Code (GitHub Copilot):

1. Open the Command Palette (Cmd+Shift+P / Ctrl+Shift+P)
2. Search for "MCP: Add Server"
3. Select "HTTP" as the server type
4. Enter the URL: https://mcp.quicknode.com/mcp
5. Choose whether to save to User Settings (global) or Workspace Settings (project-only)

Alternatively, add this to your .vscode/mcp.json:

{
  "mcp": {
    "servers": {
      "quicknode": {
        "type": "http",
        "url": "https://mcp.quicknode.com/mcp"
      }
    }
  }
}

風帆衝浪

To add the Quicknode MCP server to Windsurf:

1. Open Windsurf Settings > Cascade > MCP > Add Server > Add custom server
2. This opens ~/.codeium/windsurf/mcp_config.json. Add:

{
  "mcpServers": {
    "quicknode": {
      "serverUrl": "https://mcp.quicknode.com/mcp"
    }
  }
}

3. Click the refresh button in the MCP section of settings to activate

Zed

To add the Quicknode MCP server to Zed:

1. 開啟 Settings (Cmd+, / Ctrl+,)
2. Add the following to your settings.json:

{
  "context_servers": {
    "quicknode": {
      "transport": "http",
      "url": "https://mcp.quicknode.com/mcp"
    }
  }
}

3. Restart Zed to activate the server

Other Clients

For any other MCP-compatible client, use the following connection details:

• URL: https://mcp.quicknode.com/mcp
• Transport: Streamable HTTP
• Authentication: OAuth 2.1 (handled automatically on first connect)

If your client doesn't support OAuth or HTTP transport natively, you can use mcp-remote as a bridge:

{
  "mcpServers": {
    "quicknode": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://mcp.quicknode.com/mcp"]
    }
  }
}

Authentication and Roles

When you connect for the first time, a browser window opens where you log in with your Quicknode account. If you belong to multiple Quicknode accounts, you'll choose which one to connect. You will then select an API key role that determines what the MCP server can do on your behalf:

• Admin — Full read and write access. This allows your AI assistant to create and delete endpoints, modify security rules, update rate limits, and manage all account resources.
• Viewer — Read-only access. Your assistant can list endpoints, view metrics, logs, usage, and billing data, but cannot create, modify, or delete any resources.

This process creates an API key under the hood, scoped to the role you selected. You do not need to create or manage the key yourself.

Choosing the Right Role

Choose the Viewer role if you only need to inspect your infrastructure. Use Admin only when you need the assistant to make changes. If you need to switch roles later, remove the server from your client, re-add it, and go through the authorization flow again to re-authenticate with a different role.

Using an API Key

If you're running in a non-interactive environment where a browser-based login isn't possible (e.g., CI/CD pipelines, remote VMs, or automation agents), you can skip the OAuth flow by passing your API key as a bearer token. The MCP server recognizes the Authorization header and treats the request as already authenticated.

Generate an API key from the Quicknode Dashboard under Account (avatar icon) > API Keys. The role you assign to the key (Admin or Viewer) determines the MCP server's permissions, the same as choosing a role during the OAuth flow.

克勞德

克勞德・科德

Pass your API key as a bearer token using the --header flag:

克勞德·麥克普 新增 quicknode --傳輸 http https://mcp.quicknode.com/mcp --header "Authorization: Bearer YOUR_API_KEY"

Claude 桌面版

Add the --header flag to the mcp-remote arguments in your claude_desktop_config.json:

{
  "mcpServers": {
    "quicknode": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://mcp.quicknode.com/mcp", "--header", "Authorization: Bearer YOUR_API_KEY"]
    }
  }
}

OpenAIOpenAI

《古籍》

Pass your API key as a bearer token using the --bearer-token-env-var flag, which tells Codex to read the key from an environment variable at runtime:

codex mcp add quicknode \
  --url https://mcp.quicknode.com/mcp \
  --bearer-token-env-var YOUR_API_KEY

Run codex mcp list to confirm the server is registered.

游標

Add the following to your .cursor/mcp.json:

{
  "mcpServers": {
    "quicknode": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://mcp.quicknode.com/mcp", "--header", "Authorization: Bearer YOUR_API_KEY"]
    }
  }
}

VS Code

Add the headers field to your .vscode/mcp.json:

{
  "mcp": {
    "servers": {
      "quicknode": {
        "type": "http",
        "url": "https://mcp.quicknode.com/mcp",
        "headers": {
          "Authorization": "Bearer YOUR_API_KEY"
        }
      }
    }
  }
}

風帆衝浪

Add the headers field to your ~/.codeium/windsurf/mcp_config.json:

{
  "mcpServers": {
    "quicknode": {
      "serverUrl": "https://mcp.quicknode.com/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}

Zed

Add the headers field to your settings.json:

{
  "context_servers": {
    "quicknode": {
      "transport": "http",
      "url": "https://mcp.quicknode.com/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}

Other Clients

For any client that supports custom headers, set:

Authorization: Bearer YOUR_API_KEY

If your client uses mcp-remote, add the --header flag:

{
  "mcpServers": {
    "quicknode": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://mcp.quicknode.com/mcp", "--header", "Authorization: Bearer YOUR_API_KEY"]
    }
  }
}

Available Tools

The MCP server exposes the following tools through the Admin API. Tools marked as read-only are available to both Admin and Viewer roles. All other tools require the Admin role.

Endpoint Management

Create, read, update, and delete Quicknode endpoints across any supported chain and network.

Tool	說明	存取
list-endpoints	List all endpoints on your account	read-only
get-endpoint	Get details for a specific endpoint	read-only
create-endpoint	Create a new endpoint on a specified chain and network	管理員
delete-endpoint	Delete an endpoint	管理員
get-endpoint-metrics	Retrieve performance metrics for an endpoint	read-only
list-endpoint-logs	Fetch recent logs for an endpoint	read-only
get-endpoint-log-details	Get detailed information for a specific log entry	read-only

Rate Limits

Configure method-level and endpoint-level rate limits.

Tool	說明	存取
update-endpoint-rate-limits	Set overall rate limits (RPS, RPM, RPD) for an endpoint	管理員
list-endpoint-method-rate-limits	List method-specific rate limits	read-only
create-endpoint-method-rate-limit	Add a rate limit for a specific RPC method	管理員
update-endpoint-method-rate-limit	Update an existing method rate limit	管理員
delete-endpoint-method-rate-limit	Remove a method rate limit	管理員

安全

Manage authentication and access control settings for endpoints. The create-security-rule 以及 delete-security-rule tools support multiple rule types: IP allowlists, JWT validation, referrer restrictions, domain masks, and authentication tokens.

Tool	說明	存取
list-endpoint-security	View current security settings and rules	read-only
update-endpoint-security-options	Enable or disable security features (CORS, HSTS, IP filtering, etc.)	管理員
create-security-rule	Add a security rule (IP, JWT, referrer, domain mask, or token)	管理員
delete-security-rule	Remove a security rule	管理員

Usage and Billing

Monitor RPC usage and review billing information.

Tool	說明	存取
get-rpc-usage	Get RPC usage with optional breakdown by endpoint, method, or chain	read-only
get-billing	View billing invoices or payment history	read-only

鎖鏈

Query available chains and networks.

Tool	說明	存取
list-chains	List all supported chains and their available networks	read-only

MCP Resources

In addition to tools, the Quicknode MCP server exposes your endpoints as MCP resources. Resources let your AI assistant reference endpoint details directly in context without needing to call a tool first. In clients that support MCP resources, you can type @quicknode to browse and attach your endpoints to the conversation (e.g., quicknode://endpoints/<endpoint-id>).

This is useful when you want the assistant to have your endpoint configuration readily available while working on a task, such as writing code that targets a specific chain or debugging a request.

Example Prompts

You do not need to reference tool names directly. The AI assistant maps natural language to the appropriate tool. Some examples:

Endpoint management:

• "Create a new Base mainnet endpoint"
• "Show me all my endpoints and their current status"

Monitoring and debugging:

• "Check the logs for my Arbitrum endpoint from the last hour"
• "Show me the performance metrics for my Ethereum endpoint over the past week"

Security:

• "Add JWT authentication to my Ethereum mainnet endpoint"
• "Review security settings across all endpoints"

Usage and billing:

• "Show me my RPC usage broken down by chain for this month"
• "Which endpoint is using the most credits this month?"

Rate limiting:

• "Set a rate limit of 100 requests per second on my Polygon endpoint"
• "Add a method-specific rate limit for eth_call"

提示

• Start simple — Try asking "What endpoints do I have?" or "Show me my usage for this month" to get familiar with the available tools.
• Be specific — When you have multiple endpoints, mention the chain or name to help the assistant find the right one quickly.

How It Works

The MCP server is hosted remotely at https://mcp.quicknode.com/mcp and uses OAuth 2.1 for authentication. When your AI assistant receives a request that involves Quicknode infrastructure, it calls the appropriate MCP tool, which makes authenticated requests to the Quicknode Admin API. The assistant handles the response formatting and presents the results in natural language.

Write operations (creating endpoints, modifying security settings, deleting resources) require the Admin role and your explicit confirmation before the assistant executes them. If you connected with the Viewer role, the assistant can only perform read-only operations.

提示

Quicknode MCP manages your account-level infrastructure. If you want AI coding agents to have accurate knowledge of Quicknode's APIs and products when writing code, see Blockchain Skills. For AI-optimized documentation discovery, see LLM-Optimized Docs.

資源

Quicknode Connector on Claude

Add Quicknode directly from the Anthropic MCP directory

→

Setup Guide

Manage your blockchain infrastructure with AI and MCP

→

Quicknode on ChatGPT

Install the Quicknode app directly from the ChatGPT Apps catalog

→

管理員 API 文件

Quicknode Admin API reference

→

EVM MCP 伺服器

Guide for using the EVM MCP server

→

Solana MCP 伺服器

Guide for using the Solana MCP server

→

我們 ❤️ 您的回饋！

如果您對這份文件有任何意見或疑問，請告訴我們。我們非常樂意聽取您的意見！