What is the Neural Link?

The Neural Link is a bi-directional bridge that connects AI agents directly to the Neo.mjs runtime. It lets agents inspect the Scene Graph, component state, event listeners, computed styles, and DOM rectangles, and mutate the running application in real time.

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

Neo.mjs maintains persistent application objects in a worker-backed Scene Graph instead of compiling application state away into ephemeral DOM nodes. That architecture enables multi-window orchestration, runtime permutation, and deep AI introspection.

What is Context Engineering?

Context Engineering shapes the information and tool environment around AI agents. Neo.mjs implements it through Knowledge Base, Memory Core, GitHub Workflow, and Neural Link MCP servers for frontier harnesses, plus a File System MCP server for internal Neo.ai.Agent local loops.

What is the Neo.mjs Agent OS?

The Neo.mjs Agent OS is the repository Brain: source code and services for Memory Core, Knowledge Base, Active Hybrid GraphRAG, DreamService, Golden Path synthesis, A2A coordination, and Neural Link tooling.

Frontmatter

id	13061
title	Harness docking: define dock-zone model contract
state	Closed
labels	enhancementdesignaiarchitecture
assignees	neo-gpt
createdAt	Jun 13, 2026, 5:52 AM
updatedAt	Jun 13, 2026, 8:40 AM
githubUrl	https://github.com/neomjs/neo/issues/13061
author	neo-gpt
commentsCount	0
parentIssue	13012
subIssues	[]
subIssuesCompleted	0
subIssuesTotal	0
blockedBy	[]
blocking	[x] 13063 Define stale componentRef restore behavior for dock-zone layouts
closedAt	Jun 13, 2026, 8:40 AM

Harness docking: define dock-zone model contract

Closed v13.1.0/archive-v13-1-0-chunk-2 enhancementdesignaiarchitecture

neo-gpt commented on Jun 13, 2026, 5:52 AM

Context

Epic #13012 names the infinite-canvas / enterprise-desktop migration vector as part of the Agent Harness product line. Issue #13030 captured the market-signal and scoping anchor for QT-grade docking strategies: dock zones, nested split containers, tabbed dock panels, and save/restore layouts on top of Neo's existing multi-window and JSON blueprint substrate.

This ticket is the first concrete leaf split from #13030. It deliberately starts with the model contract, not the full UI. The prior intake comment on #13030 identified direct implementation as unsafe until the primitive boundary is chosen.

Live duplicate evidence before filing: latest 20 open issues checked at 2026-06-13T03:51:31Z; the only overlapping live issue was #13030 itself. Scope-specific search for dock-zone / docking / tabbed dock / split layout found #13030, #13028, old SortZone/grid work, and unrelated layout tickets, but no concrete dock-zone model contract leaf. Local resources/content exact sweep returned no dock-zone leaf. A2A last-50 scan showed no recent lane-claim/lane-intent on this scope; #13032 benchmark progress is separate.

The Problem

#13030 is intentionally a scoping anchor. If an implementation branch starts there directly, it risks bundling five separate concerns: semantic dock-zone model, split/tab container integration, drag-to-dock previews, blueprint persistence, and demand validation. That would make the first PR too broad and would force reviewers to decide architecture, interaction behavior, and persistence format in one cycle.

The first missing substrate is narrower: a Neo-owned model contract for what a docked layout is before any preview UI or persistence code commits to a shape.

The Architectural Reality

Existing substrate found by repo and KB checks:

Neo already has multi-window/shared-worker state and JSON blueprint persistence primitives (learn/benefits/MultiWindow.md, learn/benefits/JSONFirstUIs.md).
Neo already has layout building blocks such as vbox, hbox, and card/tab semantics (learn/guides/uibuildingblocks/Layouts.md).
Drag/drop and window choreography already exist in adjacent surfaces named by #13030, including dashboard sorting and window-manager surfaces.
The KB did not surface a dedicated docking manager or dock-zone layout primitive. That absence is the reason this leaf starts with a contract.

The model must compose with Neo primitives instead of cloning QT, Golden Layout, or Dockview. External libraries are reference semantics only.

The Fix

Define the minimal dock-zone model contract for the harness/infinite-canvas lane. The PR can be docs-only if that is the lowest-risk shape; if it introduces a new .mjs primitive, the author must run structural-pre-flight before choosing the destination.

The contract should answer:

What is the minimal tree shape for edge zones, nested splits, and tabbed dock slots?
Which layer owns the model: general layout primitive, dashboard/container primitive, or harness-specific adapter?
How does the model map onto existing vbox / hbox / card semantics without forking the drag pipeline?
Which fields are serializable as a future Neo blueprint, and which are runtime-only hover/preview state?
What is the second independent use shape beyond the anonymized QT-migration signal before implementation locks?

Contract Ledger Matrix

Target Surface	Source of Authority	Proposed Behavior	Fallback / Boundary	Docs / Evidence
Dock-zone model contract	#13030 + Epic #13012 infinite-canvas vector	Define a minimal serializable tree for edge zones, nested splits, and tabbed slots	Do not implement preview UI or persistence in this first leaf	Contract doc or JSDoc-backed code plus review evidence
Ownership boundary	Existing Neo layout/dashboard/window substrates	Decide whether the model is general layout, dashboard/container, or harness adapter	If uncertain, document the rejected alternatives and defer implementation	Cite existing repo surfaces read during the PR
Drag integration boundary	Existing drag/drop and dashboard sorting pipeline	State how future preview state will ride existing drag events	No parallel drag system in this leaf	Evidence from source reads, not external-library habit
Blueprint compatibility	Neo JSON-first / blueprint substrate	Separate serializable layout fields from runtime-only preview state	No hardcoded harness-only persistence shape	JSON example or schema-like table in the contract
Demand validation	#13030 market signal + harness self-use	Name a second independent use shape before locking the model	Partner-specific presets stay out of public repo	Body/review evidence names the second use shape generically

Decision Record impact

Aligned-with ADR 0020 (#13020). This leaf refines the harness infinite-canvas / dashboard-composition direction and does not amend an ADR.

Acceptance Criteria

A dock-zone model contract exists in a durable repo surface or JSDoc-backed code surface.
The contract covers edge zones, nested split orientation, tabbed slots, stable item identity, and serializable ordering.
The contract explicitly states the ownership layer and why the alternatives were rejected.
The contract distinguishes serializable blueprint fields from runtime-only drag/preview state.
The contract names one second independent use shape besides the anonymized QT-migration signal.
Drag-to-dock preview UI, layout persistence implementation, and split/tab rendering are not bundled into this first PR.
If new .mjs files are introduced, the PR documents structural-pre-flight evidence for the chosen path.

Out of Scope

Drag-to-dock preview visuals.
Split/tab container implementation.
Save/restore persistence implementation.
Cross-window choreography; #13025 and #13028 own that lane.
Partner-specific layout presets.

Avoided Traps

Porting an external docking library's object model: rejected. Neo needs a model that composes with its worker, layout, and blueprint substrate.
Bundling the first model contract with the UI: rejected. It would make one PR decide too many boundaries at once.
Treating #13030 as implementation-ready: rejected by prior intake evidence; this leaf is the safe first decomposition step.

Parent: Epic #13012. Scoping source: #13030. Adjacent choreography: #13025, #13028. Release classification: post-release product-line leaf; attached to Agent Harness Project 13, not the v13 release board.

Origin Session ID: 019ebe4d-0687-7dd2-8e37-c48ac8e057ad Retrieval Hint: "dock-zone model contract harness docking split tabbed slots blueprint runtime preview state"