Memmy Agent

Memmy Agent is a local memory hub for people who switch between AI assistants and want earlier decisions and project context available in the next conversation. It stores personal memory on your device by default. Direct agent tasks support your own credentials for a chosen model API. Model processing depends on the configured service or provider.

With your permission, Memmy reads supported conversation histories on your device and turns useful details into searchable memory. That can include preferences, project decisions and progress from earlier work. Connected assistants receive context relevant to the task instead of your entire conversation history.

Supported tools include Cursor, Claude Code and Codex, along with OpenClaw, Hermes, DeepSeek Harness, OpenCode and Pi. Memmy can also take tasks directly, such as combining plans from earlier conversations or drafting a document from recorded technical decisions. The memory service can run separately from the agent runtime.

You can inspect and manage stored memories, choose which conversation sources it reads, and stop further collection. Access controls let you choose which supported assistants can use the memory. Sharing context with a cloud-connected assistant doesn't keep its model requests local: Claude Code with its standard cloud connection, for example, sends prompts to its remote model provider.

The app is free and open source under MIT. Desktop builds support macOS 12 or later on Apple Silicon and Intel, and Windows 10 or later on x64. Linux has a CLI and terminal interface for x64 and ARM64, requiring Node.js 22 or newer and a systemd user session.