Compartment

What Compartment does

Compartment keeps persistent memory on your own computer, so what an agent learns in one session is available in later sessions, in other projects and to other agents on the machine. Each memory is a single claim stored with its source and the date it was learned, memories can expire after a set date, and a new preference replaces the old one. Recall is a hybrid vector and keyword search over an in-memory index, which the README says answers in about 12 ms.

Everything on disk is encrypted, including the embedding vectors, and only your passphrase opens the vault. The embedding model ships inside the package, and the README states there is no telemetry and no network use at runtime. A new vault includes about 6,700 reference facts on hardware, operating systems, ports, encodings and shell tools, which one switch removes from search. Hermes is wired in with compartment integrate hermes after pip install compartment && compartment init, and a GUI offers a memory map.

Key features

• Single encrypted vault, vectors included, opened by a passphrase
• Hybrid vector and keyword recall over an in-memory index
• Memories stored with source and date, with optional expiry
• compartment integrate command for Hermes, Claude and OpenClaw
• No API key, account, network use or telemetry
• GUI with a memory map

When to use it

• Giving Hermes Agent private long-term memory that stays on the machine
• Sharing one memory between Hermes and Claude Code
• Keeping client or personal facts encrypted at rest

Who it is for: Users who want agent memory that is local, encrypted and shared across several agents.

How it fits with Hermes Agent

Supports Hermes Agent as one of its MCP clients, with a dedicated compartment integrate hermes command.

How to install Compartment

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

pip install compartment && compartment init
compartment integrate hermes

FAQ