Nexus-Catalyst is a cross-platform memory and context layer for the AI tools you already use. One shared, searchable brain across Claude, Cursor, ChatGPT — so your context follows you instead of dying with the tab.
You finish an architecture session in Cursor. You open Claude Desktop the next morning to keep going. Everything you locked in yesterday — gone. You re‑explain. You repeat. You ship slower than you should.
Raw vector search finds memories that sound alike. A knowledge graph finds memories that are connected. NC runs both, then merges them — so recall is richer than either could be alone.
Every connected tool feeds NC the decisions, code, and direction-locks from each session. No extra ritual — just keep working.
Each memory is embedded for semantic search, and in parallel its people, projects and decisions are extracted into a knowledge graph. Similarity finds what sounds alike; the graph finds what's connected.
Start a new session anywhere. recall_context semantically searches your corpus, walks the graph, and returns the right memories — plus a persona hint for who should lead.
Instead of one generic assistant, NC routes to a specialist whose domain, communication style, and personality actually fit your task — automatically.
A decision isn't just a paragraph — it's tied to the PR that implemented it, the issue that prompted it, the thread where it landed. NC links them.
Individual use is the on-ramp. The real value is institutional: every architectural decision, every workaround, every "why we don't do it that way" — kept, searchable, and inherited by the next engineer.
Connect your first client in under five minutes. NC handles capture, embeds, and recall from the first session you run.