wazionapps/nexo

Overview

NEXO Brain provides a shared memory layer for AI agents using persistent memory, semantic RAG, and natural forgetting. It works with Claude Code, Claude Desktop, Codex, and any MCP client, offering trust scoring and a metacognitive guard. The system is fully local, open source, and free.

Best for

Best for
Developers building local, memory-enhanced multi-agent systems that need trust and forgetting

Use cases

• Give AI agents persistent cross-session memory
• Manage semantic retrieval and automatic forgetting
• Integrate trust scoring into multi-agent workflows

Notes

NEXO Brain provides a shared memory layer for AI agents using persistent memory, semantic RAG, and natural forgetting. It works with Claude Code, Claude Desktop, Codex, and any MCP client, offering trust scoring and a metacognitive guard. The system is fully local, open source, and free.

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

Use cases

• Give AI agents persistent cross-session memory
• Manage semantic retrieval and automatic forgetting
• Integrate trust scoring into multi-agent workflows

Pros

• Fully local and free, no external dependencies
• Works with multiple MCP-compatible clients
• Includes built-in trust scoring and metacognitive guard

Cons

• Low community adoption (22 stars at time of entry)
• Requires local Python environment setup
• Compatibility limited to MCP protocol clients

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