Sanity Context MCP: AI Knowledge Agent

This is a submission for the Sanity Challenge, Path One: Ship an Agent That Queries Real Content.

Project Overview

I built an AI-powered knowledge agent connected to a structured Knowledge Base stored in Sanity.

The goal of the project is to allow users to ask questions in natural language and get answers based on structured content from Sanity rather than relying only on general AI knowledge or simple keyword search.

How It Works

The application uses:

• Sanity as the structured Knowledge Base
• Sanity Context MCP to provide context from the knowledge base
• An AI agent to understand user questions and generate answers
• A web interface where users can interact with the agent

The Knowledge Base contains structured documentation and information that the agent can use when answering questions.

Key Features

• Natural-language questions
• Structured content stored in Sanity
• AI-powered answers based on Sanity content
• Source-aware responses
• Knowledge Base designed for documentation and technical information
• Web-based interface for interacting with the agent

Why Sanity Context MCP?

Instead of treating Sanity content as a simple collection of text, the project uses structured content as a knowledge source for the AI agent.

This makes it possible for the agent to retrieve relevant information from the Knowledge Base and use that information to provide more grounded answers.

Example

A user can ask:
"What is HTML?"

The agent can retrieve the relevant HTML information from the Sanity Knowledge Base and use it to generate the response. Additional knowledge entries can be added to Sanity without changing the application itself.

Sanity Project Details

• Sanity Project ID: 9m8pdh3h
• Dataset: production

Links

• GitHub Repository: https://github.com/nexus-digital-in/sanity-ai-agent
• Live Project: Local environment only (API Integration WIP)
• Sanity Studio: [https://www.sanity.io/@oj12djnwi/studio/ea7r3qhu66e1oyi9u193zjpm/default/structure]

Tech Stack

• React
• Vite
• Sanity
• Sanity Context MCP
• AI API
• Cloudflare Workers
• JavaScript

Project Goal

The main goal of this project was to explore how Sanity's structured content can be connected to an AI agent and used as a reliable knowledge source. This project is also designed so the Knowledge Base can be expanded with more technical documentation and structured information in the future.

Challenge

This project was created as part of the Sanity Challenge, Path One, to demonstrate an agent that can query real, structured content and use it to answer user questions.

Thanks to the Sanity team and Dev Community for organizing the challenge.