Memmy Agent is a local-first personal AI runtime and shared memory hub for multiple assistants. It collects and structures knowledge, preferences, project history, and prior conversations so tools such as Claude Code, Codex, Cursor, OpenClaw, and Hermes Agent can reuse the same context. Historical sessions can be imported and converted into searchable long-term memory instead of remaining isolated in separate applications. The runtime supports a desktop app, an interactive CLI, a terminal interface, memory commands, and an OpenAI-compatible API. Skills, MCP servers, browser tools, messaging channels, and productivity integrations extend it from conversation into task execution. Users can choose account-based access or configure their own models and API keys. Memory, configuration, and application state remain on the local machine by default.
Features
- Shared long-term memory across AI agents
- Historical conversation and project import
- Desktop, CLI, TUI, and API access
- MCP, Skills, and managed browser tools
- Messaging and productivity platform integrations
- Local-first storage with flexible model configuration