Hermes Atlas
Memory & knowledge · works with Hermes Agent

Memmy

MemTensor/memmy-agent

Local memory hub and personal agent that shares one memory across Claude Code, Codex, OpenClaw and Hermes

In short

Memmy is a personal AI agent and local memory service that gives several coding agents one shared, locally stored memory. Hermes Agent is listed among the agents that can connect to it, alongside Claude Code, Codex and OpenClaw.

What Memmy does

Memmy ships as a desktop app, a memmy CLI and TUI, a standalone memmy-memory CLI, and a source build. The memory service listens on http://127.0.0.1:18960 by default, and memmy-memory commands initialize it, check health, and search, add and get memories. Running memmy-memory init installs the Memory Skill and the supported hook or plugin for each detected agent, or for one agent chosen with --agent.

The CLI path targets Linux x64 or arm64 with a systemd user session. The installer enables a local memory service and, on first run, a gateway service, both bound to localhost. memmy serve starts an OpenAI-compatible API on port 18990, and providers are set in ~/.memmy/config.yaml with a bring-your-own-key option. The roadmap in the README mentions adding memory sources beyond AI conversations, such as browser activity and local documents, and agent-to-agent collaboration for teams.

Key features

  • Shared local memory store used by Claude Code, Codex, Cursor, OpenClaw, Hermes and other agents
  • memmy-memory CLI with init, health, search, add and get commands
  • Desktop app, interactive TUI and a single-turn memmy agent command
  • OpenAI-compatible API started with memmy serve
  • Bring-your-own-key mode configured in ~/.memmy/config.yaml
  • systemd user services for the memory and gateway components on Linux

When to use it

  • Continuing the same project across Claude Code, Codex and Hermes without re-explaining context
  • Giving scripts a searchable local memory through memmy-memory search and add
  • Running a personal agent from a desktop app on trial credits, then switching to your own model API

Who it is for: Developers who switch between several coding agents and want a single local memory that all of them read and write.

How it fits with Hermes Agent

Memmy lists Hermes among the agents it can connect to and the repository carries the hermes-agent topic; it is a general memory hub, not a Hermes-only plugin.

How to install Memmy

These commands are copied from the project's README. Check the repository for the latest steps before you run them.

curl -fsSL https://raw.githubusercontent.com/MemTensor/memmy-agent/main/scripts/install.sh | bash
memmy

Requirements: For the CLI: Linux x64 or arm64, Node.js 22 or newer and an available systemd user session; the desktop app is downloaded from memmy.bot or GitHub Releases

Note: Registration grants trial tokens for the bundled agent runtime, and once they run out you must supply your own model API key.

FAQ

What is Memmy?

Memmy is a personal AI agent and local memory hub. It keeps one shared memory on your machine that several AI agents can read from and write to, so context carries between them.

Does Memmy work with Hermes Agent?

Yes, the README lists Hermes among the agents that can connect to Memmy. Run memmy-memory init to install the Memory Skill and supported hook or plugin for detected agents, or add --agent to pick one.

Is Memmy free and open source?

The code is MIT licensed. The bundled agent runtime starts with trial credits from registration and then expects your own model API key.

Similar memory for Hermes Agent

All memory

Related guides: SOUL.md for Hermes Agent: what it is and how to write one · Run multiple Hermes agents with profiles