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	11367
title	Land ADR 0004: GitHub Content Architecture (universal ordinal-100 chunking)
state	Closed
labels	documentationenhancementaiagent-task:pendingarchitecturemodel-experience
assignees	neo-opus-4-7
createdAt	May 14, 2026, 6:55 PM
updatedAt	May 14, 2026, 9:10 PM
githubUrl	https://github.com/neomjs/neo/issues/11367
author	neo-opus-4-7
commentsCount	0
parentIssue	null
subIssues	[]
subIssuesCompleted	0
subIssuesTotal	0
blockedBy	[]
blocking	[]
closedAt	May 14, 2026, 9:10 PM

Land ADR 0004: GitHub Content Architecture (universal ordinal-100 chunking)

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

Closeddocumentationenhancementaiagent-task:pendingarchitecturemodel-experience

neo-opus-4-7 commented on May 14, 2026, 6:55 PM

Context

Operator-directed 2026-05-14 (verbatim): "what i would strongly recommend is a ticket for a new decision record as the new single source of truth (future us will be grateful)."

This ticket lands learn/agentos/decisions/0004-github-content-architecture.md as the authoritative single source of truth for the on-disk shape of resources/content/. The ADR captures the universal ordinal-100 chunking primitive confirmed by operator on 2026-05-14 — Reading X (full symmetry across active + archive + all 3 GitHub types + release-notes).

This is a docs-only landing. The migration / config audit / consumer rewires / LocalFileService index-map substrate / chunkPath.mjs retirement all ride as separate downstream tickets filed AFTER this ADR graduates and provides authority for those work products.

The Problem

PR #11362 (commit 559c73d43, 2026-05-14) deleted 3,366 archived items as "legacy" instead of reshaping them per Epic #11187 Phase 3 ACs. Root cause: substrate-bypass — graduation context lived across multiple Discussions / Epic body amendments / memory citations; author inferred deletion semantics from operator-canonicalized phrases without V-B-A'ing against the explicit Phase 3 reshape mandate.

Operator framing 2026-05-14: "create decision records, store tons of input into MC, and yet fail to apply the GRADUATED architectures." The substrate intelligence existed; the failure was at execution-time discipline (not consulting the existing graduated substrate before authoring).

This ADR solves the load-bearing-substrate-anchor problem: future agents have ONE document to V-B-A against, instead of re-inferring from multi-Discussion graduation contexts that are easy to misread in isolation.

The Architectural Reality

Two prior chunking primitives lived in code:

ai/services/github-workflow/shared/chunkPath.mjs — active-tier ID-range (<NNN>xx/)
ai/services/github-workflow/shared/archivePath.mjs — archive-tier ordinal-100 (chunk-N/)

Operator-confirmed direction (Reading X, 2026-05-14): retire ID-range entirely; universal ordinal-100 everywhere. Quote: "no symmetric per tier. consistency is key. same for all. own id chunks. no here we use A, there we use B. clean architecture design."

Affected files (for this ticket's docs-commit ONLY):

learn/agentos/decisions/0004-github-content-architecture.md — NEW file

Affected files (downstream tickets, NOT this ticket's scope):

ai/services/github-workflow/shared/chunkPath.mjs — retire
ai/services/github-workflow/shared/archivePath.mjs — consolidate into universal helper
ai/services/github-workflow/sync/{IssueSyncer,PullRequestSyncer,DiscussionSyncer}.mjs — call universal helper
ai/services/github-workflow/LocalFileService.mjs — index-map based lookup
ai/mcp/server/github-workflow/config{,.template}.mjs — drop historical-debt configurability
resources/content/{issues,pulls,discussions,release-notes,archive}/* — delete + resync via sync_all
All consumers (TicketSource, PullRequestSource, DiscussionSource, IssueIngestor, ticket-index, SEO routes) — recursive walk + index lookup

The Fix

Single-commit PR landing learn/agentos/decisions/0004-github-content-architecture.md. Content is the full ADR draft staged at /tmp/0004-github-content-architecture.md for operator review pre-commit. Sections:

Context (the friction, the substrate-bypass anti-pattern anchor)
Decision (target shape, universal ordinal-100 rule, retired primitives, prevent-reopen.yml foundation, insertion-order semantics)
Implementation Details (universal helper proposal, index-map substrate, syncer write-paths, release-cut, consumer recursion, recovery shape, config audit)
Consequences (positive + negative honestly listed)
Anti-Patterns (5 specific failure modes including #11362's substrate-bypass)
V-B-A Pre-Flight for Future Authors (mandatory reads before authoring resources/content/ migrations)
Related
Status / Lifecycle
Downstream Tickets (the migration + config + consumer work, scoped OUT of this ticket)

ADR follows existing learn/agentos/decisions/000{1,2,3}-*.md precedent (header table + numbered sections + status lifecycle).

Acceptance Criteria

learn/agentos/decisions/0004-github-content-architecture.md committed and merged to dev
ADR Status flipped from "Draft" → "Accepted" before merge (operator-approved content)
PR body cites this ticket via Closes #N
PR has 2-peer cross-family review (per pull-request §6.1 cross-family mandate)
Per pr-review §8.2 Cross-Skill Integration Audit: verify no other skill / doc needs to reference the new ADR as predecessor (likely TBD — possibly ticket-create-workflow.md should reference ADR 0004 for migration-class tickets)

Out of Scope

Migration execution (delete + resync via sync_all) — separate downstream ticket
Universal helper (contentPath.mjs) implementation — separate downstream ticket
Index-map substrate — separate downstream ticket
LocalFileService rewrite — separate downstream ticket
Config audit + drop historical-debt configurability — separate downstream ticket
Release-notes chunking + new ReleaseNotesSyncer — separate downstream ticket
publish.mjs review — separate downstream ticket
Consumer rewires — separate downstream ticket
Recovery of #11362-deleted archives (3,366 items) — separate downstream ticket, depends on universal-helper landing

Avoided Traps

Bundle migration into the ADR commit — rejected; ADR is authority, migration is implementation, separating them lets each PR review focus
Skip the ADR and just file migration ticket directly — rejected; that's exactly the substrate-bypass pattern (no single-source-of-truth anchor; future agents re-derive)
Sub-issue under Epic #11187 — considered; rejected because the ADR SUPERSEDES Epic #11187's Cycle 2 amendment, so it's an authority-shift not a sub-task. ADR cites Epic #11187 in Related; Epic body should reference ADR 0004 as Cycle 3 amendment after merge
Use Ideation Sandbox first — considered; rejected because architecture is operator-confirmed (Reading X) and substrate is empirically V-B-A'd convergent. Ideation Sandbox is for unsettled questions; the question is settled, only the codification needs landing
Treat as "documentation-only / low-priority" — rejected; this is high-load-bearing substrate that gates downstream substrate-mutation work. Land first, build forward

Epic #11187 — current shape with Cycle 2 amendment; this ADR is the Cycle 3 codification (operator-confirmed Reading X)
Discussion #11180 — parent ideation; Option G superseded by symmetric direction
Discussion #11359 — Phase 6 graduation that triggered #11362 substrate-bypass
PR #11362 — the substrate-bypass failure this ADR anchors against
PR #11193 — introduced archivePath() partial precursor primitive
PR #11114, #11123, #11125, #11129 — introduced chunkPath() (now retired)
.github/workflows/prevent-reopen.yml — load-bearing immutability primitive
learn/agentos/decisions/0003-chroma-topology-unified-only.md — sibling ADR for pattern reference

Origin Session ID

cf76b29a-9cf5-4c35-a415-37d631a8a755

Handoff Retrieval Hints

query_raw_memories("github content architecture ADR universal ordinal chunk 100 prevent-reopen")
Operator quote anchor: "a ticket for a new decision record as the new single source of truth (future us will be grateful)" (2026-05-14)
Commit-range anchor: 8a1906221..559c73d43 for substrate-bypass evidence
Staged ADR file path: /tmp/0004-github-content-architecture.md (Claude session-local; survives only until session sunset)

tobiu referenced in commit d8ba634 - "docs(agentos): land ADR 0004 GitHub content architecture (#11367) (#11368) on May 14, 2026, 9:10 PM

tobiu closed this issue on May 14, 2026, 9:10 PM