Local Memory

Acerca de

This tool functions as a local Model Context Protocol (MCP) server, specifically designed to empower AI and coding agents with durable, searchable memory. It allows agents to store, retrieve, update, and delete operational context, decisions, and insights across chat sessions, ensuring that valuable information like API contracts or project-specific gotchas are never lost. By running entirely locally with an embedded vector database (Zvec) and local embedding models (Ollama), it offers a private and efficient solution for agents to maintain semantic understanding and adapt their behavior over time without relying on external services.

Características Principales

• Workspace-scoped memory management for multiple projects
• Utilizes Ollama for local embedding generation (e.g., embeddinggemma)
• Semantic search over stored memories using an embedded vector database
• Store, supersede, and delete memory entries for AI agents
• Runs entirely locally, eliminating the need for external vector databases
• 4 GitHub stars

Casos de Uso

• Maintaining durable coding context across AI agent chat sessions
• Retrieving relevant past context semantically for AI agent responses
• Correcting and refining an AI agent's memory by superseding or deleting outdated entries