FSRS Memory provides a robust, local-first solution for managing AI long-term memory, designed to store, recall, and review contextual information efficiently. It leverages FSRS (forgetting-space repetition scheduling) to prioritize memories likely to be forgotten, ensuring AI agents receive concise, high-priority context and reducing API token costs. With git-backed snapshots, semantic diffing, and project isolation, it offers a reliable system for developers and applications to manage their knowledge bases, including features for soft-delete, bulk operations, and optional summarization via Anthropic.
主な機能
01Soft-delete, restore, and purge workflow for memory management
020 GitHub stars
03Optional Anthropic summarization for project memories
04FSRS-based memory review scheduling for optimal recall
05Git-backed automatic backups with semantic diffing and restore tools
06Local-first memory storage with support for isolated projects
