soolaugust/0CompactMem

Overview

0CompactMem provides persistent memory for Claude Code and other LLM agents using OS memory primitives like demand paging and mlock pinning. It stores memory in a single SQLite file and is MCP-native, allowing multiple agents to share the same memory.

Best for

Best for
Developers building multi-agent LLM systems that need efficient, shared persistent memory with OS-level control

Use cases

• Persistent memory for LLM agent sessions across restarts
• Shared memory context for multi-agent collaboration
• Efficient memory management using OS-level demand paging

How to use

Install

pip install -e .

Tools exposed

• MCP-native

Tested with

Claude Code, Cursor

Notes

0CompactMem provides persistent memory for Claude Code and other LLM agents using OS memory primitives like demand paging and mlock pinning. It stores memory in a single SQLite file and is MCP-native, allowing multiple agents to share the same memory.

0 stars on GitHub. Last updated 2026-06-01.

Use cases

• Persistent memory for LLM agent sessions across restarts
• Shared memory context for multi-agent collaboration
• Efficient memory management using OS-level demand paging

Pros

• Leverages OS memory management (demand paging, mlock) for performance
• Single SQLite file simplifies storage and portability
• MCP-native design enables seamless multi-agent memory sharing

Cons

• Low community adoption (0 stars) suggests early-stage or niche tool
• Requires system-level configuration of OS primitives (swap, mlock)
• Potential complexity in tuning memory constraints for different workloads

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.