What is the 'Neural Link'?

The Neural Link is a bi-directional bridge that connects AI agents (like Gemini or Claude) directly to the Neo.mjs runtime. It allows agents to 'see' the application's Scene Graph, inspect component state, verify event listeners, and even mutate the running application in real-time. This turns the application into a 'glass box' for AI, enabling autonomous debugging and feature development.

Why is Neo.mjs called an 'Application Engine' instead of a framework?

Traditional frameworks are general-purpose libraries (like a Toyota) that help you organize code, but they compile away into ephemeral DOM nodes. Neo.mjs is a precision-engineered runtime (like an F1 car), similar to Unreal Engine for games. It maintains a persistent 'Scene Graph' of objects in a separate worker thread. These objects retain their identity, state, and relationships, allowing for advanced capabilities like multi-window orchestration, runtime permutation, and deep AI introspection that are impossible with 'melted plastic' DOM-based frameworks.

What is 'Context Engineering'?

Context Engineering is the practice of curating the environment and information flow for AI agents to maximize their autonomy. In Neo.mjs, this is implemented via three Model Context Protocol (MCP) servers: a Knowledge Base for semantic code understanding, a Memory Core for learning from past sessions, and a GitHub Workflow server for project management. This ecosystem allows agents to work as fully integrated members of the development team.

What is 'Object Permanence' in the context of Neo.mjs?

In Neo.mjs, UI components are persistent JavaScript objects living in the App Worker, not just transient rendering results. This 'Object Permanence' means a component (like a Dashboard) maintains its state (scroll position, user input, internal logic) even if it is detached from the DOM or moved to a different browser window. This is the 'Lego Technic' approach versus the 'Duplo' approach of traditional frameworks.

What is an 'Agent Operating System'?

Neo.mjs v11 introduced the concept of an Agent OS, where the platform itself provides the tools and interfaces for AI agents to operate. It combines a standalone, type-safe AI SDK for autonomous 'Code Execution' with the Neural Link for runtime control. This enables agents to monitor, debug, and heal the application autonomously, effectively acting as an operating system for synthetic intelligence.

How does Neo.mjs handle multi-window applications?

Neo.mjs uses shared web workers to run a single application instance across multiple browser windows. All windows share the same application state and data in real-time. Components can even move between windows while retaining their JavaScript instances. This enables desktop-class experiences like multi-window IDEs, browser-based email clients, and multi-screen control rooms.

What makes Neo.mjs different from React, Angular, or Vue?

Neo.mjs is fundamentally different in its architecture: (1) It uses True Multithreading via web workers to prevent UI jank, (2) It is an AI-Native Application Engine with built-in bridges for AI collaboration, and (3) It treats components as persistent objects in a Scene Graph, enabling native multi-window support and runtime permutation.

Frontmatter

id	9768
title	Refactor memory-core Database Lifecycle & Vector Managers
state	Closed
labels	airefactoringarchitecture
assignees	tobiu
createdAt	Apr 7, 2026, 11:45 PM
updatedAt	Apr 8, 2026, 1:03 AM
githubUrl	https://github.com/neomjs/neo/issues/9768
author	tobiu
commentsCount	1
parentIssue	null
subIssues	[]
subIssuesCompleted	0
subIssuesTotal	0
blockedBy	[]
blocking	[]
closedAt	Apr 8, 2026, 1:03 AM

Refactor memory-core Database Lifecycle & Vector Managers

Name: Neo.mjs Application Engine
Author: Neo.mjs

Closedairefactoringarchitecture

tobiu commented on Apr 7, 2026, 11:45 PM

This epic covers the architectural teardown and modularization of the DatabaseLifecycleService to support configuration-driven Vector Databases (Chroma vs SQLite) and local Inference Daemons (openAiCompatible via MLX).

Architectural Context

The memory-core MCP server's initial infrastructure outgrew its design:

The Vector Sprawl: DatabaseLifecycleService organically grew into a "god class" that intertwines Chroma and local inference logic, making it impossible to run cleanly if an engine was disabled.
Missing Mac Provisioning: The recent switch to openAiCompatible left Apple Silicon environments without an automated provisioning path for the MLX daemon (the Ollama equivalent).
The Proxy Coupling: CollectionProxy was statically importing mutually exclusive managers (ChromaManager and SQLiteVectorManager), breaking the dependency inversion principles established by aiConfig.engine.

Implementation Plan

A finalized implementation plan has been written to address this:

MLX Provisioning: Create a setup_mlx.sh script to explicitly pull gemma4:31b and qwen3-embedding via python mlx_lm and run the daemon on port 11435 to avoid collisions.
Vector Clean Architecture: Implement an AbstractVectorManager and use a dynamic factory inside CollectionProxy to load managers.
Service Refactoring: Dismantle DatabaseLifecycleService into ChromaLifecycleService and InferenceLifecycleService, properly exporting them via ai/services.mjs using the singleton .ready() paradigm.

Note: This issue was created to transfer context to a fresh AI session.

tobiu added the ai label on Apr 7, 2026, 11:45 PM

tobiu added the refactoring label on Apr 7, 2026, 11:45 PM

tobiu added the architecture label on Apr 7, 2026, 11:45 PM

tobiu assigned to @tobiu on Apr 7, 2026, 11:47 PM

tobiu Apr 8, 2026, 12:07 AM

Input from Gemini 3.1 Pro (Antigravity):

✦ The Memory Core architecture has been refactored in this session.

Vector Managers: The CollectionProxy now dynamically imports vector managers (Chroma or SQLite) based on aiConfig.engine. Both ChromaManager and SQLiteVectorManager now correctly inherit from a shared AbstractVectorManager.

Lifecycle Modularization: The monolithic DatabaseLifecycleService was dismantled.

Explicit Daemons: We introduced ChromaLifecycleService dedicated purely to ChromaDB orchestration, and InferenceLifecycleService for local inference daemons (Ollama / MLX).

Verification: The memory-core MCP server was manually verified to boot correctly, waiting asynchronously for both the Inference and Chroma services before resolving transport.

This concludes the architectural teardown. The system is structurally robust against config deviations.

tobiu referenced in commit c8da337 - "refactor(ai): modularize memory-core vector managers & lifecycles (#9768) on Apr 8, 2026, 12:10 AM

tobiu referenced in commit 403d9e7 - "fix(ai): resolve ollama auto-boot enoent on apple silicon (#9768) on Apr 8, 2026, 12:14 AM

tobiu referenced in commit 4d7386b - "fix(ai): enhance offline ollama probe to support windows (#9768) on Apr 8, 2026, 12:15 AM

tobiu referenced in commit 33b3b88 - "docs(ai): apply Knowledge Base Enhancement Strategy to new abstraction layers (#9768) on Apr 8, 2026, 12:19 AM

tobiu closed this issue on Apr 8, 2026, 1:03 AM