MCP support in NeetoKB

Imagine you could just ask for the article you published last week, search your entire knowledge base for a topic, draft a new help doc, or invite a teammate — all without opening NeetoKB, clicking through folders, or copying and pasting links.

NeetoKB works with AI assistants you may already be using, like Claude, Cursor, and GitHub Copilot. Once you connect them together, your AI assistant can read your articles, search your knowledge base, create and update content, and manage team members. You simply ask in plain English and the assistant does the work.

What is MCP?

MCP stands for Model Context Protocol. It is an open standard that lets AI assistants and apps like NeetoKB speak the same language. You do not need to understand how it works to use any of the steps below — think of it as the plumbing that connects your AI assistant to your NeetoKB workspace.

Why this is useful

Think of it as giving your AI a passcode to your NeetoKB workspace. After that, the usual back-and-forth disappears. A few examples of what becomes possible:

• You ask, "What articles do we have about billing?" and the assistant searches your knowledge base.

• You paste a public article link and ask, "Summarize this help article for me."

• You say, "Create a draft article in the Getting Started folder about resetting your password."

• You ask, "List all published articles in the API folder."

• You say, "Invite [email protected] to the workspace."

You do not need to be technical to benefit from this. If you already talk to an AI assistant for other things, you can talk to it about your NeetoKB content too.

What you need to get started

Two things:

1. An AI assistant that supports MCP-style server connections.

2. A NeetoKB API key, which authenticates the assistant with your workspace.
   See the API Keys help article for how to generate one.

Once you have both, you paste a small settings snippet into your AI tool and restart it.

Connecting your AI assistant to NeetoKB

Pick the tool you use. You only need to follow the steps for that one; you can skip the rest. Each section shows a small block of settings — copy it, replace YOUR_API_KEY with your own key, and save.

Claude Code

Open the file ~/.claude.json and paste the following (if the file already has settings, add just the neetokb entry inside mcpServers):

{
    "mcpServers": {
      "neetokb": {
        "type": "http",
        "url": "https://connect.neetokb.com/mcp/messages",
        "headers": {
          "Authorization": "Bearer YOUR_API_KEY"
        }
      }
    }
  }

Save the file and restart Claude Code.

Cursor

Open ~/.cursor/mcp.json and paste the following:

{
    "mcpServers": {
      "neetokb": {
        "type": "http",
        "url": "https://connect.neetokb.com/mcp/messages",
        "headers": {
          "Authorization": "Bearer YOUR_API_KEY"
        }
      }
    }
}

Save it and restart Cursor (or reload the window via the command palette).

VS Code with GitHub Copilot

Create or open .vscode/mcp.json in your workspace and paste:

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

VS Code will ask you for your API key the first time. You need VS Code 1.99 or newer, and you must be in the Copilot chat's Agent mode to use it.

Windsurf

Open ~/.codeium/windsurf/mcp_config.json and paste:

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

Save it and make sure the connection is turned on in Settings > Cascade > MCP Servers. Windsurf has a limit of 100 total tools across all connected services, so if it complains, turn a few other connections off.

Try it out

Once you are connected, the easiest way to see what is possible is to just ask. Here are some prompts you can copy and paste to get a feel for it:

• "List all published articles in my knowledge base."

• "Search my KB for articles about password reset."

• "Get the full content of the article with slug api-keys."

• "Show me the category tree for my knowledge base."

• "What is my knowledge base called and what is its public URL?"

• "Create a draft article titled 'Reset your password' in the Account category."

• "Update the billing article to published state."

• "List active team members in my workspace."

• "Summarize the public article at https://help.neetokb.com/articles/mcp."

What your AI assistant can actually do

Under the hood, the connection gives your AI assistant a set of abilities. You do not need to remember any of these names — just ask naturally and the assistant picks the right one. But here is the full list if you are curious.

• Finding articles: list articles with filters, search published articles by keyword (with highlighted snippets), and fetch a single article including drafts and full HTML content.

• Creating and updating articles: create new articles with a category path that auto-creates missing folders, and update an article's title, HTML body, slug, or publish state.

• Organizing and browsing: list categories as a hierarchical tree with article counts, list article authors, and list article recommendations.

• Workspace settings: fetch your knowledge base name and public root URL.

• Team members: list active team members, fetch a member by ID, invite members by email, update a member's details or role, and deactivate a member.

There is also a read-only public ability for any shared NeetoKB article link from another workspace — published, publicly accessible content only. Your connected workspace is where create, update, and manage actions apply.