Learn the Permaweb with Permamind [Guest Post]

Permamind Header

Introduction

What is Permamind?
- While Permamind is the world’s first permanent, decentralized AI memory system built on Arweave and AO, this guide focuses on its power as your personal, always-current documentation assistant for the Permaweb ecosystem
- It’s pre-loaded with the latest official documentation (AO Cookbook, HyperBEAM, etc.) and acts as an informed LLM that can answer complex technical questions accurately
- Think of it as having a Permaweb expert available 24/7 who never gives outdated or hallucinated answers
What This Guide Covers
- This post will walk you through the entire setup process, from installing the server to connecting your favorite AI client and running your first query
- We will provide specific instructions for Claude Desktop, Cursor, and Raycast
Why This Matters
- Instead of digging through multiple docs or getting generic/outdated responses from standard LLMs, you’ll have instant access to accurate, current information about the Permaweb tech stack
- Perfect for developers building on AO, learning the Permaweb, or beginning their journey with HyperBEAM

Part 1: Installing the Permamind Server

Prerequisites

You’ll need Node.js version 20 or higher. You can check your version by running:
```
node -v
```

Installation Steps

To install permamind, open your terminal and use npm to install the Permamind package globally.
```
npm i -g permamind
```

Part 2: Connecting Your AI Client

Now that your local Permamind server is running, let’s connect it to your AI-powered development tool.

Claude Desktop

Locate your Claude Desktop config file:
- macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
- Windows: %APPDATA%\Claude\claude_desktop_config.json

Add the Permamind server configuration: Open the config file and add the following MCP server configuration:

{
  "mcpServers": {
    "Permamind": {
      "command": "npx",
      "args": ["permamind"]
    }
  }
}

Restart Claude Desktop for the changes to take effect.
Verify the connection: You should see the Permamind server listed in your available tools/servers within Claude Desktop.

Cursor

Open your Cursor editor.
Go to Cursor Settings > Tools & Integrations.
Click New MCP server.

Add the following configuration:

{
  "mcpServers": {
    "Permamind": {
      "command": "npx",
      "args": ["permamind"]
    }
  }
}

Save the configuration and restart Cursor for the changes to take effect.

Raycast

Open Raycast and search for “MCP” or navigate to AI settings.
Search for Install MCP Server.
Fill in the configuration form:

Field Value
Transport Select Standard Input/Output
Command Enter npx
Arguments Enter permamind
Description Custom description of this server
Cmd + Return to install the server.
Invoke the MCP server using @permamind <Your query here>

Field	Value
Transport	Select `Standard Input/Output`
Command	Enter `npx`
Arguments	Enter `permamind`
Description	Custom description of this server

Part 3: A Sample Run

Let’s confirm everything is working. The Permamind server you’re running has been pre-configured with the latest documentation for AO, the Permaweb Cookbook, HyperBEAM, and more.

The Prompt

In your configured client (Claude, Cursor, or Raycast), start a new chat and ask a question that requires knowledge of the AO ecosystem.

For example:

What is HyperBEAM?
How can I expose my AO process to HyperBEAM?
How can I use AR.IO’s Wayfinder SDK?

Expected Output

Your AI client should provide a detailed, accurate answer based on the official documentation. It won’t give a generic or hallucinated response because Permamind is feeding it the correct context.

Example Expose AO Process

How does it work?

Your client sent the prompt to your local Permamind server. The server identified the topic and used the integrated Permaweb docs to provide the LLM with the necessary information to form a correct response.

Conclusion & Next Steps

Congratulations! You now have a personal, always-current permaweb assistant running locally. You’ve successfully installed the server, connected a client, and verified its ability to provide accurate, up-to-date answers using our ecosystem’s official documentation.

This is a huge boost to your development workflow, enabling you to learn, build, and debug with Permaweb technologies faster than ever - no more outdated answers or generic LLM responses when working with AO, HyperBEAM, or any part of the Permaweb.

About the authors

Dylan Shade is a software developer at Forward Research. Follow Dylan on X.

Jonathon Green is a software developer exploring the Arweave/AO ecosystem. Follow Jonathan on X.

Disclosure: This is a guest post submitted by external contributors. The views expressed are their own and do not necessarily reflect those of Permaweb Journal.

Permaweb Journal