AI Integration

ZineCore2 publishes its documentation in machine-readable formats designed for AI tools and large language models. This means AI assistants can understand the ZineCore2 specification, help you write valid metadata, and query controlled vocabularies directly.

llms.txt

The site serves an llms.txt file at /llms.txt, following the emerging standard for making website content accessible to LLMs. This file provides a structured index of all ZineCore2 documentation, including profile specifications, vocabulary definitions, and usage guides.

A full-content version is also available at /llms-full.txt, which includes the complete text of every documentation page in a single file. This is useful for loading the entire ZineCore2 specification into an AI assistant's context window.

What's included

• Profile specifications for all four profiles (ZineCore2, AgentCore2, HoldingCore2, RepoCore2)
• Controlled vocabulary documentation (subjects, genres, rights, agent roles, etc.)
• Tool documentation (validator, playground, API reference)
• Standards references and namespace definitions

How to use it

Point any LLM or AI coding assistant at the llms.txt URL to give it full knowledge of the ZineCore2 specification:

https://zinecore.org/llms.txt
https://zinecore.org/llms-full.txt

Many AI tools support llms.txt natively. You can also paste the contents into a chat session or system prompt to give an AI assistant full context about ZineCore2.

MCP Server

The site also exposes a Model Context Protocol (MCP) server, allowing AI tools like Claude, Cursor, and VS Code Copilot to interact with ZineCore2 documentation programmatically.

What MCP provides

MCP goes beyond static text by giving AI tools structured access to:

• Documentation lookup - Query specific profile definitions, element semantics, and usage examples
• Navigation - Browse the documentation tree and find relevant pages
• Live data - Access the same content that powers this website

Connecting to the MCP server

Add the ZineCore2 MCP server to your AI tool's configuration:

Claude Desktop (claude_desktop_config.json):

{
  "mcpServers": {
    "zinecore": {
      "url": "https://zinecore.org/mcp"
    }
  }
}

VS Code (.vscode/mcp.json):

{
  "servers": {
    "zinecore": {
      "url": "https://zinecore.org/mcp"
    }
  }
}

Cursor (.cursor/mcp.json):

{
  "mcpServers": {
    "zinecore": {
      "url": "https://zinecore.org/mcp"
    }
  }
}

Claude Code CLI:

claude mcp add --transport http ZineCore2 https://zinecore.org/mcp

Use cases

• Cataloging assistance - Ask an AI to help you fill out a ZineCore2 record with the correct fields and vocabulary terms
• Schema validation - Have an AI check your JSON records against the specification
• Code generation - Generate models, forms, or API endpoints that conform to ZineCore2 profiles
• Vocabulary lookup - Query subject terms, genres, or rights statements without leaving your editor

Combining both approaches

For the best experience, use both:

• llms.txt gives AI tools a broad understanding of the entire specification upfront
• MCP lets AI tools look up specific details on demand as questions arise

Together, they enable AI assistants to provide accurate, specification-aware help when working with zine metadata.