Skip to main content
Most products have far more code than documentation. When your docs are sparse, outdated, or scattered across forgotten wikis, the Product Graph MCP can analyze your codebase to build a structured Product Artifact hierarchy: The MCP connects directly to your IDE or AI assistant, letting you discover capabilities from code structure, extract features from routes and components, and generate documentation from your codebase itself.

Prerequisites

1

Set up MCP integration

Follow the MCP Integration guide to connect Product Graph to your IDE or AI assistant.
2

Access your codebase

Ensure your AI assistant can read your codebase. In Cursor or similar tools, simply open your project.

Mapping Code to Artifacts

Best Practices

  • Start small: Begin with one capability, validate the workflow, then expand
  • Review AI output: The AI may combine concepts that should be separate or miss edge cases
  • Get approval first: Ask the AI to summarize proposed changes before executing
  • Iterate: Create structure first with basic content, then enrich over time

Workflow

Each prompt follows the same workflow:
1

Understand the Template

The AI retrieves the artifact template from Product Graph to learn what information is required.
2

Discover

The AI analyzes your codebase for relevant information based on the artifact type.
3

Pull Existing Data

The AI fetches any existing artifacts from Product Graph to avoid duplicates.
4

Reconcile

The AI creates a matrix comparing what it found in code vs. what exists in Product Graph, identifying additions, updates, and removals.
5

Approval

You review the reconciliation matrix and approve before any changes are made.

Prompts

Product

Capabilities

Features

Requirements

Acceptance Criteria

Core Entities

User Types

Next Steps

MCP Integration

Full reference for all MCP tools and configuration options.

Product Artifacts

Deep dive into each artifact type and best practices.