Octonove/crbro-memory

Overview

CRBRO is a persistent neural memory system for AI agents, implemented as an MCP server. It stores memory in a file-based structure with neurons, synapses, and heat scores that persist across sessions. It works with Claude Code and any MCP client.

Best for

Best for
Developers building persistent memory for MCP-based AI agents

Use cases

• Give AI agents long-term memory across chat sessions
• Store and retrieve knowledge in a file-based neural graph
• Integrate persistent memory into Claude Code or MCP-compatible tools

How to use

Install

npx crbro-memory init

Tools exposed

• crbro_boot
• crbro_inspect
• crbro_learn
• crbro_recall
• crbro_revise
• crbro_forget
• crbro_connect
• crbro_context
• crbro_map
• crbro_consolidate
• crbro_maintenance
• crbro_audit
• crbro_secret
• crbro_space
• crbro_share
• crbro_status
• crbro_neuron
• crbro_neurons
• crbro_hot_topics
• crbro_connections

Tested with

Claude Desktop, Claude Code, Cursor, Windsurf

Notes

CRBRO is a persistent neural memory system for AI agents, implemented as an MCP server. It stores memory in a file-based structure with neurons, synapses, and heat scores that persist across sessions. It works with Claude Code and any MCP client.

5 stars on GitHub. Last updated 2026-09-12. Licensed MIT.

Use cases

• Give AI agents long-term memory across chat sessions
• Store and retrieve knowledge in a file-based neural graph
• Integrate persistent memory into Claude Code or MCP-compatible tools

Pros

• Persistent memory survives across sessions
• File-based design is simple and portable
• Works with any MCP client

Cons

• Very early stage with minimal adoption (5 stars)
• File-based storage may not scale to large memory graphs
• Limited documentation or ecosystem beyond the repo

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