Contextberg captures and exposes the contextual signals from a user's work (the "context beneath your screen") and turns that history into structured, accessible memory for AI agents. It exposes those memory artifacts over the Model Context Protocol (MCP), enabling agents and developer tools to request, consume, and reuse long-term context across sessions. By standardizing how work-derived context is served, Contextberg reduces context-switching, helps agents make decisions with richer historical information, and enables interoperable memory across agent ecosystems.