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	9471
title	Add Anti-Corruption Protocol to AGENTS.md
state	Closed
labels	documentationai
assignees	tobiu
createdAt	Mar 14, 2026, 12:06 PM
updatedAt	Mar 15, 2026, 7:37 PM
githubUrl	https://github.com/neomjs/neo/issues/9471
author	tobiu
commentsCount	3
parentIssue	null
subIssues	[]
subIssuesCompleted	0
subIssuesTotal	0
blockedBy	[]
blocking	[]
closedAt	Mar 14, 2026, 12:11 PM

Add Anti-Corruption Protocol to AGENTS.md

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

Closed v12.1.0 documentationai

tobiu commented on Mar 14, 2026, 12:06 PM

Description

The AI agents have been experiencing "Contextual Drift" (context corruption) during long, complex sessions. When an agent modifies a file multiple times and then performs a full read_file to "refresh" its understanding, the context window fills with competing semantic realities of the same file. This confuses the attention mechanism, leading to hallucinated string replacements and failed edits.

Additionally, agents have been falling into a "Context Blindness" trap, using inefficient shell commands (cat and grep) to search files they already have in context, wasting tokens and causing severe API delays.

Proposed Solution

Add a new Section 8 to AGENTS.md titled "Preventing Context Corruption (State Management)".

This protocol will:

Enforce Single Reads: Restrict agents to reading a file fully only once per session.
Ban Re-reads: Explicitly forbid re-reading modified files to prevent competing realities.
Mandate git diff: Instruct agents to use git diff HEAD to reconcile state when they lose track of edits, anchoring their understanding without duplicating code.
Enforce grep_search: Mandate the use of the native grep_search tool for targeted method verification instead of full file reads.
Ban Shell Fallbacks: Strictly forbid the use of cat or grep via run_shell_command for file reading/searching.

tobiu added the documentation label on Mar 14, 2026, 12:06 PM

tobiu added the ai label on Mar 14, 2026, 12:06 PM

tobiu referenced in commit c8e951f - "docs: Add Anti-Corruption Protocol to AGENTS.md (#9471)" on Mar 14, 2026, 12:10 PM

tobiu Mar 14, 2026, 12:10 PM

Input from Gemini 3.1 Pro:

✦ The new "Anti-Corruption Protocol" has been successfully added to AGENTS.md. This will instruct all future AI sessions to avoid context drift by relying on git diff and targeted grep_search instead of dangerous full-file re-reads and inefficient cat/grep shell commands.

tobiu assigned to @tobiu on Mar 14, 2026, 12:10 PM

tobiu closed this issue on Mar 14, 2026, 12:11 PM

tobiu Mar 15, 2026, 7:37 PM

@raye-deng Thanks for your input! The git diff idea sounds good for edits. I am still mostly using Gemini CLI with the 3.1 pro model. I did notice that the CLI instructions recently changed, and not in a good way. Starting with

CRITICAL INSTRUCTION 1: Use specific tools.

The intent is good, but it seems to just confuse models. The changes of this ticket definitely need more refinements: now Gemini is frequently using the Search_Text tool inside known files.

What helped me the most to counter hallucinations is this one: https://github.com/neomjs/neo/blob/dev/learn/guides/mcp/NeuralLink.md

Best regards, Tobi