If your AI agents keep losing track of what happened last week or which teammate knows what, this gives them organizational memory. It ingests events from your tools via REST or gRPC, builds a four-layer knowledge graph (session, daily, project, org) using PostgreSQL and Apache AGE, then serves compressed context summaries in around 250 tokens. The MCP server exposes context retrieval so Claude Desktop or Cursor can pull relevant history, decisions, and expertise without you pasting chat logs. Best for teams running multiple AI workflows that need to share knowledge across sessions and projects.
claude mcp add --transport http dotnova-context-layer https://api.dotnova.io/mcp