Memmy

Memmy is an open-source personal AI agent and memory layer designed to preserve context across otherwise separate AI tools. It can examine authorized local conversation histories, organize useful details into searchable memory, and supply relevant context when work moves between agents. A local agent runtime also lets users execute tasks directly instead of treating Memmy solely as an archive.

Highlights

• Shares project goals, preferences, decisions, and prior attempts across supported AI agents
• Imports authorized histories from tools including Codex, Claude Code, Cursor, and OpenCode
• Stores memory and application state locally, with controls for access, inspection, and deletion
• Provides desktop, CLI/TUI, memory-service, and OpenAI-compatible API entry points
• Extends workflows through MCP, Skills, messaging channels, browser tools, and third-party integrations

Memmy is best suited to developers, technical teams, and intensive AI users who frequently alternate between coding agents or need durable project context. The software is free and open source, can be built and operated locally, and supports bring-your-own-key model configuration. Account mode may include limited trial credits, while ongoing model usage can be paid through the provider selected by the user.