Contextualize is an experimental application for designing, organizing, and reusing context across projects. Rather than treating conversations, notes, documentation, and resources as isolated artifacts, it explores how knowledge can become a connected visual workspace that supports both human thinking and AI collaboration.
This page is a guided walkthrough of the project. It introduces the application, documents the tools and workflows used to build it, and provides direct access to the live prototype, technical resources, early designs, and the methodology behind its development.
https://vimeo.com/1214462440?share=copy&fl=sv&fe=ci
Current state: A deployed prototype and visual canvas are available to explore. The context model, MCP interface, permissions, and AI-assisted workflows remain experimental and are still being tested.
Contextualize explores MCP in two related ways. During development, Codex connected to Lovable and Supabase to help work across the application stack. As a product experiment, Contextualize is also being explored as an MCP server so an AI assistant can use selected context actions without requiring an embedded chat interface.
Only explicitly published tools are available to an assistant. The current tool surface, access model, and connection flow should be treated as experimental.
https://docs.lovable.dev/features/agent-integrations
The documentation above explains Lovable’s app-to-assistant connection model. The captured setup below records how that model is being explored in this prototype. The next section follows the corresponding build workflow.
The resources below trace the implementation from the live canvas to the platforms and agent workflow behind it. Start with the prototype, then use the remaining links to understand the contribution of each tool.