Practical overview
When it fits
Use it when an MCP-connected agent needs persistent, locally stored context across sessions or projects, rather than a hosted memory service.
Who it is for
- Developers configuring MCP-compatible AI assistants.
- Users who want agent context to persist across sessions and projects on one computer.
- People evaluating local encrypted storage for agent memory.
Documented uses
- Keep useful agent context available after a conversation ends.
- Share one persistent memory vault among MCP clients on the same machine.
- Search stored memories using vector and keyword retrieval.
- Track the source and date of a memory and set an expiry date.
Requirements & limits
The README describes local operation without an account or API key. It documents installation with pip and initialization using the compartment command.
- Requires a Python-based local installation and an MCP client to use agent read/write access.
- The embedding model is included in the package; the README describes hybrid vector and keyword search.
- The vault and embedding vectors are described as encrypted, with access opened by a passphrase.
- The project documents a graphical panel and memory map alongside command-line tools.
- The repository specifies the Apache-2.0 license.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Install Compartment and connect it to Claude
Create a local Compartment vault and register it as Claude’s memory through MCP.
Before you start
- Python package installation using pip.
- Claude Code or Claude Desktop.
- A passphrase for initializing the vault.
Steps
Install Compartment with the command below.
pip install compartmentCreate the vault with
compartment initand provide a passphrase when prompted.Run
compartment integrate claudeto register Compartment with Claude and apply the documented Claude integration.
Expected outcome
Claude can use Compartment’s MCP memory tools, including search, store, recent, forget, status, and lock.
Guide limitations
- This integration guide covers Claude Code and Claude Desktop; other clients have separate setup paths.
- The integration imports existing Claude Code memory files without modifying them; removing those files is a separate manual action.
- The source describes the vault as encrypted and local, with no API key or account required.
Install Compartment and select it as Hermes Agent memory
Set up Compartment and select it through Hermes Agent’s memory provider picker.
Before you start
- Python package installation using pip.
- Hermes Agent.
Steps
Install Compartment with the command below.
pip install compartmentCreate a vault with
compartment init.Run
compartment integrate hermesto configure the Hermes integration.In Hermes, use
hermes memory setupand select Compartment from the provider picker.
Expected outcome
Hermes is configured to use Compartment as its memory provider; the documented integration exposes search, store, and forget tools.
Guide limitations
- The source says the vault stays unlocked until reboot after initialization; it does not document a different unlock workflow for this setup.
- The source describes Compartment as a local encrypted provider that needs no API key or cloud account.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 14 October 2026
- Original website title
- MaxFreedomPollard/Compartment · Official homepage title · retained in its source language