AI Tool

nexo-brain Review

NEXO Brain is an open-source cognitive runtime providing a shared brain for AI agents with persistent memory and durable workflows.

nexo-brain - AI tool
1Operates 100% locally on CPU, requiring no GPU or cloud dependencies, and is free/libre under the AGPL-3.0 license.
2Achieved over 10,000 monthly downloads on npm by March-April 2026, with a peak of 1,281 downloads in a single day.
3Version 0.3.1 (March 2026) introduced over 97 MCP tools across 18 categories, alongside Prediction Error Gating and a Security Pipeline.
4Demonstrated an F1 score of 0.588 on LoCoMo (ACL 2024), representing a 55% improvement over GPT-4 (128K), with 93.3% adversarial rejection.

nexo-brain at a Glance

Best For
AI developers and researchers
Pricing
Subscription SaaS
Key Features
Shared-brain memory, Goal Engine, Decision Cortex, Outcome learning, Skill evolution
Integrations
See website
Alternatives
CrewAI, Engram, Zep, Cognee
🏢

About nexo-brain

Business Model
Subscription SaaS
Target Audience
AI developers and researchers

Similar Tools

Compare Alternatives

Other tools you might consider

4

LLC Memory Server

Shares tags: ai

Visit

Connect

</>Embed "Featured on Stork" Badge
Badge previewBadge preview light
<a href="https://www.stork.ai/en/nexo-brain" target="_blank" rel="noopener noreferrer"><img src="https://www.stork.ai/api/badge/nexo-brain?style=dark" alt="nexo-brain - Featured on Stork.ai" height="36" /></a>
[![nexo-brain - Featured on Stork.ai](https://www.stork.ai/api/badge/nexo-brain?style=dark)](https://www.stork.ai/en/nexo-brain)

overview

What is nexo-brain?

nexo-brain is a cognitive runtime tool developed by NEXO Brain that enables AI operators, developers, and small teams to provide persistent memory and durable workflows for AI agents. It functions as a local cognitive runtime, modeling human memory with sensory register, short-term memory, and long-term memory, along with Ebbinghaus forgetting curves. This system transforms AI agents from stateless assistants into persistent, learning, and adaptive cognitive partners, specifically designed for AI coding agents like Claude Code, Codex, and Gemini CLI.

quick facts

Quick Facts

AttributeValue
DeveloperNEXO Brain
Business ModelOpen Source / Freemium
PricingFree
PlatformsLocal (CPU only), compatible with Claude Code, Codex, Claude Desktop, any MCP client
API AvailableNo
IntegrationsClaude Code, Codex, Claude Desktop, any MCP client

features

Key Features of nexo-brain

NEXO Brain incorporates a human-inspired cognitive architecture, directly modeling the Atkinson-Shiffrin multi-store memory model and Ebbinghaus forgetting curves. It includes a sensory register, working memory, and long-term memory, with rehearsal-based strengthening and memory consolidation during 'sleep' processes. The system is maintained by autonomous background processes, including 9 background processes (part of its 'nervous system' of 14 cron jobs) that ensure memory health, generate intelligence briefs, and self-repair without user intervention.

  • 1Shared-brain memory for AI agents, enabling collaborative intelligence across multiple clients.
  • 2Persistent memory across sessions, eliminating the need to re-explain context to AI agents.
  • 3Semantic RAG (Retrieval Augmented Generation) for efficient and relevant information retrieval.
  • 4Natural forgetting mechanisms based on Ebbinghaus curves to manage memory decay and consolidation.
  • 5Metacognitive guard and Cognitive Dissonance Detection for self-healing and error avoidance.
  • 6Trust scoring (0-100) to help agents understand intent and adjust behavior based on feedback.
  • 7Over 150 MCP (Metacognitive Processing) tools across 18 categories for diverse operational tasks.
  • 8Goal Engine and Decision Cortex for facilitating complex operations and workflow understanding.
  • 9Deep Sleep pipeline for overnight learning, memory consolidation, and adaptation.
  • 10Autonomous background processes, including a daily Ebbinghaus memory sweep, for continuous system maintenance.

use cases

Who Should Use nexo-brain?

NEXO Brain is primarily designed for AI operators, developers, and small teams seeking to enhance the capabilities of their AI agents. It addresses the pain point of stateless AI agents by providing a robust, persistent, and adaptive cognitive system. The tool is particularly beneficial for those working with AI coding agents and requiring durable workflows and advanced cognitive functionalities.

  • 1AI operators: For providing persistent memory for AI agents across sessions and enabling durable workflow execution and configurable automation.
  • 2Developers: For building AI systems with advanced cognitive capabilities, including metacognition, self-healing, and human-inspired memory models.
  • 3Small teams: To enable a shared cognitive brain across multiple AI clients like Claude Code and Codex, fostering collaborative AI environments.
  • 4AI developers and researchers: For implementing overnight learning and adaptation through a 'Deep Sleep' pipeline and exploring autonomous background processes.

pricing

nexo-brain Pricing & Plans

NEXO Brain is free/libre and open-source software, released under the AGPL-3.0 license. It operates 100% locally on the user's machine, ensuring zero API costs and no cloud dependencies. All data is stored in local SQLite databases, and vector embeddings are computed on-device using ONNX Runtime on the CPU, eliminating the need for a GPU. This model provides full functionality without any associated costs.

  • 1Free: Full functionality, 100% local operation, AGPL-3.0 license, zero API costs, no cloud dependencies.

competitors

nexo-brain vs Competitors

NEXO Brain differentiates itself in the market by offering a comprehensive, self-managing local cognitive runtime that integrates persistent memory, durable workflows, and advanced metacognitive capabilities. Unlike many frameworks that require assembly, nexo-brain provides a pre-wired 'shared brain' solution.

1
LlamaIndex

LlamaIndex provides a data framework for LLM applications, enabling them to ingest, structure, and access private or domain-specific data for RAG.

While not exclusively an 'AI agent brain,' LlamaIndex offers robust RAG capabilities and persistent memory management, which are core components of nexo-brain. It's a framework rather than a ready-to-use shared brain, requiring more development effort, and its local-first emphasis depends on the user's implementation choices, unlike nexo-brain's explicit 100% local claim.

2
LangChain

LangChain is a comprehensive framework for developing applications powered by language models, offering modules for agents, memory, and RAG.

LangChain provides the building blocks for creating AI agents with persistent memory and RAG, similar to nexo-brain's underlying functionalities. However, it's a development framework, requiring users to assemble components, whereas nexo-brain presents itself as a more integrated 'shared brain' solution. LangChain's local operation depends on the integrated LLM and vector store, while nexo-brain explicitly states 100% local.

3
MemGPT

MemGPT is an open-source operating system for LLMs, designed to give them self-editing memory and the ability to manage their own context.

MemGPT directly addresses the persistent memory and context management challenges for AI agents, aligning closely with nexo-brain's 'persistent memory' and 'natural forgetting' features. It focuses heavily on the memory aspect for individual agents, potentially offering a more granular control over memory than a shared brain, but might require additional orchestration for multi-agent collaboration compared to nexo-brain's 'shared brain' concept.

4
Open-Assistant (OASST)

Open-Assistant is a chat-based AI that can interact with third-party tools and APIs, aiming to be a free and open-source alternative to commercial assistants.

While Open-Assistant is more of a conversational AI, its ability to interact with tools and APIs touches upon the '150+ MCP tools' aspect of nexo-brain. It's open-source and free, similar to nexo-brain, but its primary focus is on being a helpful assistant rather than a backend 'shared brain' for other AI agents, making it a less direct competitor in terms of core functionality.

Frequently Asked Questions

+What is nexo-brain?

nexo-brain is a cognitive runtime tool developed by NEXO Brain that enables AI operators, developers, and small teams to provide persistent memory and durable workflows for AI agents. It functions as a local cognitive runtime, modeling human memory with sensory register, short-term memory, and long-term memory, along with Ebbinghaus forgetting curves.

+Is nexo-brain free?

Yes, nexo-brain is free/libre and open-source software, released under the AGPL-3.0 license. It operates 100% locally on the user's machine, incurring zero API costs and requiring no cloud dependencies.

+What are the main features of nexo-brain?

Key features of nexo-brain include shared-brain memory for AI agents, persistent memory across sessions, semantic RAG, natural forgetting based on Ebbinghaus curves, a metacognitive guard, trust scoring, over 150 MCP tools, a Goal Engine, Decision Cortex, and a Deep Sleep pipeline for overnight learning and consolidation.

+Who should use nexo-brain?

Nexo-brain is intended for AI operators, developers, and small teams. It is particularly useful for those who need to provide persistent memory for AI agents, enable shared cognitive brains across multiple AI clients like Claude Code and Codex, facilitate durable workflow execution, and implement advanced metacognitive capabilities in AI systems.

+How does nexo-brain compare to alternatives?

Nexo-brain differentiates itself by offering a comprehensive, 100% local, and pre-assembled shared cognitive runtime. Unlike frameworks like LlamaIndex and LangChain, which provide building blocks, nexo-brain offers an integrated solution. Compared to MemGPT, which focuses on individual LLM memory, nexo-brain emphasizes a shared brain for multiple agents. It also provides a richer cognitive runtime than simpler memory solutions or conversational AIs like Open-Assistant.