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	10646
title	Require live open-queue sweep in ticket-create
state	Open
labels	documentationenhancementaiarchitecture
assignees	[]
createdAt	May 3, 2026, 4:36 PM
updatedAt	May 3, 2026, 5:02 PM
githubUrl	https://github.com/neomjs/neo/issues/10646
author	neo-gpt
commentsCount	1
parentIssue	null
subIssues	[]
subIssuesCompleted	0
subIssuesTotal	0
blockedBy	[]
blocking	[]

Require live open-queue sweep in ticket-create

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

Opendocumentationenhancementaiarchitecture

neo-gpt commented on May 3, 2026, 4:36 PM

Context

During follow-up ticket creation on 2026-05-03, @tobiu surfaced a live creation-side gap:

our ask tool can be stale. a dup sweep MUST include to at least read the latest 20 open ticket titles, authors and labels.

This correction matters immediately. ask_knowledge_base(type: 'ticket') and local resources/content/issues/** are useful, but both can lag behind GitHub reality. A ticket-create duplicate sweep that only uses stale indexed/local state can still file a near-duplicate against work opened minutes earlier by another swarm agent.

I applied the stricter rule before filing this ticket: the latest 20 open queue was fetched live from GitHub with number, title, author, labels, and URL. No open ticket in that list covered this exact ticket-create live-open-queue mandate.

The Problem

.agents/skills/ticket-create/references/ticket-create-workflow.md §1a currently makes ask_knowledge_base(type: 'ticket') the primary semantic duplicate search and local grep/query fallback the exact-match layer. That is insufficient under active swarm conditions because:

the Knowledge Base can be stale after recent issue creation or sync lag;
local resources/content/issues/** can be stale for the same reason;
multiple agents can create tickets minutes apart while operating from different harnesses/clones;
the most likely duplicates are often in the newest open queue, before sync/indexing catches up.

The result is a false sense of duplicate-sweep safety: the agent performs a sweep, but the sweep is not live enough to protect the current tracker.

The Architectural Reality

Relevant protocol surfaces:

AGENTS.md §6 Meta Gate: Deduplication & Linking — currently mandates local duplicate sweeps before issue/discussion creation.
.agents/skills/ticket-create/references/ticket-create-workflow.md §1a — creation-side Content Sweep, currently KB-primary plus local exact-match fallback.
.agents/skills/ticket-create/SKILL.md — lightweight Progressive Disclosure router; per create-skill, this should stay small and point at the reference payload.
.agents/skills/create-skill/references/skill-authoring-guide.md — confirms operational lessons should be promoted into the relevant skill payload, not bloated into the router.

Adjacent but not duplicate:

#10276 adds adjacency sweep discipline to ideation-sandbox; this ticket is specifically about issue creation and live open-ticket freshness.
#10104 created the ticket-create skill itself; it is closed and does not contain the latest-20 live queue rule.

The Fix

Update the ticket creation protocol so every create_issue path must include a live open-queue sweep before filing.

Concrete prescription:

In .agents/skills/ticket-create/references/ticket-create-workflow.md §1a, add a mandatory live GitHub sweep step:

gh issue list --state open --limit 20 --json number,title,author,labels,url

or the equivalent GitHub MCP/API call returning the same fields.

State that this live sweep is additive, not a replacement:
- keep semantic search (ask_knowledge_base(type: 'ticket')) for conceptual adjacency;
- keep local rg/grep over issues, archives, and discussions for exact historical checks;
- require the live latest-20 open queue because KB/local sync can be stale.
Add failure handling:
- if the live latest-20 call fails due sandbox/network/auth, retry with the appropriate approved/escalated path;
- if it still cannot be fetched, stop before create_issue and report the blocker rather than filing from stale evidence only.
Update AGENTS.md §6 so the per-turn invariant also names live latest-open queue freshness, not only local grep.
Keep SKILL.md unchanged unless the trigger wording itself needs a tiny pointer. The heavy rule belongs in references/ticket-create-workflow.md per Progressive Disclosure.

Acceptance Criteria

ticket-create-workflow.md §1a requires reading at least the latest 20 open issue titles, authors, labels, issue numbers, and URLs from live GitHub before create_issue.
The workflow says the live latest-20 sweep is mandatory even when ask_knowledge_base and local grep return no duplicates.
The workflow documents retry/escalation behavior for sandbox/network/auth failures and forbids proceeding to create_issue if live latest-open state cannot be fetched.
AGENTS.md §6 Meta Gate is updated to include the same live latest-open queue requirement for issue creation.
The rule is placed in the reference payload, not expanded into ticket-create/SKILL.md, preserving Progressive Disclosure.
The ticket body / creation notes pattern is updated to mention the live-open sweep result when filing future Fat Tickets.

Out of Scope

Replacing semantic duplicate search with live title scanning. Both are needed.
Requiring exhaustive full-repo remote issue enumeration on every ticket. Latest 20 open is the minimum freshness guard; deeper searches remain keyword/semantic targeted.
Changing ideation-sandbox adjacency sweep behavior; #10276 owns that lane.
Building a new MCP tool solely for this workflow. Existing GitHub API / gh issue list / workflow MCP equivalents are sufficient.

Avoided Traps

Trap: trust ask_knowledge_base because it is semantically richer. Rejected. It can be stale, and freshness is the failure mode this rule addresses.
Trap: trust local synced markdown. Rejected. resources/content/** can lag behind remote GitHub during active swarm work.
Trap: dump the rule into SKILL.md. Rejected. SKILL.md is a lightweight router; the heavy operational rule belongs in the reference payload.
Trap: make latest-20 the only duplicate sweep. Rejected. Newest open issues catch freshness; semantic/local sweeps catch older conceptual duplicates and archived/discussion adjacency.

#10104 — created the ticket-create skill.
#10276 — analogous adjacency sweep gap for ideation-sandbox.
#10645 — ticket created immediately before this correction; future tickets should include the live-open sweep explicitly.

Origin Session ID: 89b259c3-27ec-4afb-baaf-fd39b55bffe1

Retrieval Hint: ticket-create duplicate sweep latest 20 open issues titles authors labels stale ask_knowledge_base.