multi-memory

A Hermes memory provider that runs multiple memory backends at once.

---

Install

hermes plugins install someaka/multi-memory-plugin
hermes config set memory.provider multi

Then pick your backends with the interactive setup wizard:

hermes multi setup

Or add backends directly:

hermes multi add holographic
hermes multi add mnemosyne

Restart Hermes. The provider auto-discovers installed backends.

---

How it works

Hermes normally only lets you use one memory system at a time. This plugin lets you use several at once.

You tell it which ones you want. It talks to all of them whenever Hermes needs to remember something or look something up. The answers come back combined — as if they all came from one place.

If one of them stops working, the error is logged immediately. The other backends keep working fine.

---

CLI

Available after restarting Hermes:

hermes multi setup            # interactive curses-based setup wizard
hermes multi setup <name>     # configure a specific backend interactively
hermes multi status           # active backends + health + plugin status
hermes multi list             # all backends, active markers
hermes multi add <name>       # add a backend to config
hermes multi remove <name>    # remove a backend from config

The setup wizard walks through per-backend configuration (API keys, model choices, endpoint URLs) and auto-installs Python dependencies from each backend's plugin.yaml.

---

Docs

• CONFIG.md — config formats, per-backend reference, troubleshooting
• CONTRIBUTING.md — setup, tests, adding backends, architecture
• AGENT.md — instructions for AI coding assistants
• CHANGELOG.md — version history

License

AGPL-3.0-or-later