MaxFreedomPollard/Compartment

Encrypted, fully offline agentic memory. One click install, GUI w/ memory map, all OS and agents. Superior memory creation, storage and retrieval.

What it solves

Compartment provides a secure, fully offline memory store for AI agents running on a personal computer. It ensures that anything an agent learns—decisions, preferences, facts—persists across sessions without sending data to the cloud, requiring API keys, or exposing it to telemetry.

How it works

The tool creates a single encrypted vault file on the host machine. Agents write memories as short claims (≤200 chars) with source, date, and optional expiry. An embedded 384‑dimensional int8 ONNX model generates vectors for each claim; these vectors and a keyword index are kept in RAM for fast hybrid search (≈12 ms). The vault encrypts both raw text and vectors, unlocked only with a user passphrase. A lightweight GUI/app shows vault status, connected agents, and recent memories, while a local‑only dashboard visualises growth, tags, and relations.

Who it’s for

Developers and power users who run local LLMs or AI agents (e.g., Claude Code, Hermes, OpenClaw, Cursor) and want their agents to retain knowledge across runs without any network dependency or external service. It also suits privacy‑focused users who need encrypted, offline storage of personal or operational facts.

Highlights

  • End‑to‑end offline guarantee: no network, no API keys, no telemetry.
  • Encrypted single‑file vault with passphrase protection; vectors are encrypted too.
  • Hybrid vector‑keyword search with relevance scoring in ~12 ms.
  • Automatic expiry, opinion replacement, and importance weighting for memories.
  • Integrated hooks for popular agents (Claude, Hermes, OpenClaw) and one‑click installation for VS Code, Cursor, LM Studio, etc.
  • Visual dashboard (local‑only) for browsing, tagging, and relation graphs.
  • Bundled embedding model runs in ~70 MB process; no LLM required for indexing.

Related

  • Project
    doobidoo/mcp-memory-service

    A persistent, shared memory backend for AI agents that provides a knowledge graph and semantic search across sessions and tools via REST API and MCP.

  • Project
    rohitg00/agentmemory

    A persistent memory server for AI coding agents that automatically captures and retrieves project context across sessions using hybrid search.

  • Project
    mem9-ai/mem9

    A persistent memory layer for AI agents that enables shared context and recall across sessions, machines, and different agent runtimes.

  • Project
    JuliusBrussee/cavemem

    A cross-agent persistent memory system for coding assistants that uses compressed local storage and MCP to allow AI agents to remember information across sessions.

  • Project
    sopaco/cortex-mem

    A Rust-based long-term memory framework for AI agents that uses a three-tier hierarchical architecture and a virtual filesystem to provide persistent, context-aware interactions.