Self-evolving memory OS for LLM & AI Agents: ultra-persistent memory, hybrid-retrieval, and cross-task skill reuse, with 35.24% token savings and DeepSeek Harness support.
Co-manage memory with your coding agent across sessions. A local workbench for Claude Code, Codex and Pi.