Nexus is a knowledge-first product that ingests a user's saved material (bookmarks, articles, X/Twitter, Reddit saves, LinkedIn posts, notes, docs etc) and turns those scattered items into a cited, navigable, agent-ready knowledge base. It focuses on the retrieval gap — not capture — by extracting claims and concepts, auto-indexing content into book-like structures (chapters, sections, index paths), and providing cited semantic search and distilled answers that point back to sources. Nexus advertises multi-source ingestion, a self-organising taxonomy, a filesystem-backed knowledge store, and an agentic extraction pipeline to convert saves into reusable context rather than an unread queue. The product is agent-native: it exposes a standards-based MCP endpoint (https://mcp.usenexus.in/mcp) and supports chat, CLI, and MCP access so agents and tools can request cited material programmatically.