aiqbee-ai

Aiqbee's MCP server provides access to a comprehensive set of AI-assisted capabilities that connect your Aiqbee knowledge graphs, brains, and neurons to external AI assistants via the Model Context Protocol. After connecting, you can search, retrieve, and manage brains and neurons, create relationships, manage access controls, and leverage built-in tools for brain discovery, neuron management, relationships, and access control. The system supports OAuth 2.0 authentication for both public (PKCE) and confidential (server-side PKCE) clients, so you can sign in with Claude Desktop, Cursor, ChatGPT, or other MCP-compatible clients and begin issuing context-aware queries and commands through your chosen interface. Typical workflows include listing brains, creating neurons, linking relationships, and updating brain templates directly from natural language prompts or structured prompts in your client interface. In addition to core data management, there are 25 available tools spanning brain discovery, neuron management, relationship handling, and access control, all reachable through the MCP endpoint.

To use the server, configure your MCP client to point at https://mcp.aiqbee.com/mcp. Depending on your client, you may use the npx transport for stdio-based clients, or configure a JSON-based server entry in your application's MCP settings. When you connect for the first time, you’ll be prompted to sign in with your Aiqbee account via OAuth 2.0. Public clients (like Claude Desktop or Cursor) will complete the standard PKCE flow, while confidential clients (such as ChatGPT) will use server-side PKCE with a callback URL as part of the authorization process. After authentication, you can begin issuing prompts such as listing brains, creating a new brain from a template, or querying neuron content across your knowledge graph.

If you’re using an IDE or CLI tool, you can interact with the server’s capabilities by selecting the aiqbee MCP server configuration and issuing prompts like: list brains, show brain templates, create neuron, or link two neurons. The system is designed to support both structured MCP commands and natural-language prompts, enabling seamless integration into your existing AI-assisted workflows.

Prerequisites:

• Access to an MCP-compatible client (e.g., Claude Desktop, Cursor, VS Code with MCP extension, npx for stdio clients).
• OAuth 2.0 support enabled by your Aiqbee account.
• Network access to https://mcp.aiqbee.com/mcp.

Installation / setup steps:

1. Ensure you have a compatible MCP client installed. For stdio-based clients, you can use npx:

# Example for stdio MCP clients
npx -y @aiqbee/mcp

2. Configure the client to point at the Aiqbee MCP server endpoint:

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

3. Authenticate when prompted. The first connection will trigger OAuth 2.0 sign-in via your Aiqbee account. Public clients use the standard PKCE flow; confidential clients use server-side PKCE with a callback.

4. (Optional) If your client requires environment variables, there are none required by default. Ensure your network allows outbound connections to mcp.aiqbee.com.

5. Start issuing prompts or commands through your client’s MCP integration to manage brains, neurons, relationships, and access controls.

Notes:

• If you are configuring a GUI client (Claude Desktop, Cursor, etc.), use the provided MCP server URL in the client’s MCP configuration UI.
• For npx-based or command-line workflows, ensure you are using the correct package name (@aiqbee/mcp) and that you accept any prompts for authentication during first run.

Tips and common considerations:

• The MCP server uses OAuth 2.0, so you won’t need API keys. Sign in with your existing Aiqbee account.
• The server exposes 25 tools across Brain Discovery, Search & Retrieval, Neuron management, Relationships, and Access Control. Leverage these tools to automate governance and knowledge graph operations from your MCP-compatible client.
• If you encounter permission errors, verify that you have access to the target brain and associated resources in Aiqbee.
• When working with VS Code or other IDE integrations, you can add the MCP server under the servers section with type 'sse' and URL https://mcp.aiqbee.com/mcp.
• If your client requires specific transport settings (e.g., SSE vs. standard REST polling), follow the client’s MCP integration documentation to choose the appropriate transport.
• For self-hosted workflows, this endpoint is the recommended public MCP endpoint; there is no local self-hosted server configuration provided in this repository.