Claude Total Memory

关于

Claude Total Memory addresses the fundamental limitation of Large Language Models forgetting context between sessions by acting as a persistent memory MCP server for Claude Code and OpenAI Codex CLI. It empowers developers with a self-improving agent that recalls architectural decisions, past solutions, and learned lessons across sessions, preventing repetitive context provision and common mistakes. This tool offers an advanced 4-tier search pipeline, a comprehensive set of 20 MCP tools, a web dashboard, and robust privacy features, all managed through a local, private database without reliance on external cloud services or API keys.

主要功能

• Automatic privacy stripping redacts sensitive data before storage
• 4-tier intelligent search pipeline (FTS5, semantic, fuzzy, graph)
• Web dashboard for statistics, knowledge visualization, and live activity feed
• 3 GitHub stars
• Self-improving agent learns from errors and promotes insights to behavioral rules (SOUL)
• Persistent local database with knowledge graph and version history

使用案例

• Preventing AI agents like Claude from repeatedly asking for project context across development sessions.
• Building a cumulative, searchable knowledge base of solutions, decisions, and lessons for AI coding assistants.
• Enabling AI agents to learn from past mistakes and improve their performance over time.