Iranti is an open, self-hosted Model Context Protocol (MCP) server designed to provide persistent, consistent, and identity-based memory for multi-agent AI systems, such as Claude, Codex CLI, and GitHub Copilot. It enables these agents and tools to share and retrieve exact facts, checkpoints, and semantic search data across sessions and tools, reducing repetitive context loss and 'forgetting' between workflows. It includes sophisticated staff agents for conflict resolution, drift detection, session recovery, and automated memory management, with deep integration into leading AI developer tools.
Visit Iranti's official website for product details and getting started.