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	15562
title	Land rejected AgentCard directions as design evidence
state	Closed
labels	enhancementdesignai
assignees	neo-gpt
createdAt	Jul 19, 2026, 8:15 AM
updatedAt	Jul 19, 2026, 9:16 PM
githubUrl	https://github.com/neomjs/neo/issues/15562
author	neo-gpt
commentsCount	0
parentIssue	null
subIssues	[]
subIssuesCompleted	0
subIssuesTotal	0
contentTrust
projected
quarantined	0
signals	[]
blockedBy	[]
blocking	[]
closedAt	Jul 19, 2026, 9:16 PM

Land rejected AgentCard directions as design evidence

Closed Backlog/active-chunk-8 enhancementdesignai

neo-gpt commented on Jul 19, 2026, 8:15 AM

Context

Fold ticket for PR #15547 (lane-fold of #15536). The operator selected evolved D/synthesis on 2026-07-19 (selection receipt); A and B/C therefore become the rejected-direction record, not a competing production implementation. The already-complete comparison PR cannot honestly flip from draft to ready while it only carries Refs #15536: operator rule #12367 requires one real Resolves #N target on ready agent PRs, while the upcoming AgentCard implementation—not this record—resolves the parent design ticket.

The fold convention was proven by sibling ticket #15561 / PR #15538. This ticket gives PR #15547 an honest merge authority so its comparison receipts can land on dev without misrepresenting them as the selected implementation. Screenshots are optional per the operator's 2026-07-19 verdict.

Live latest-open sweep: checked the newest 20 open issues immediately before creation, plus the recent A2A in-flight claim window, ticket KB, and repository issue/discussion corpus; no equivalent rejected-direction fold ticket existed.

The Problem

PR #15547 preserves two comparison artifacts at exact head bde617a320:

apps/agentos/design/agentcard-rebaseline-a.html — A's roomy scan-column direction.
apps/agentos/design/agentcard-rebaseline-bc.html — the B/C identity-first fusion.

Euclid continued Grace's stranded draft by repairing the 294/360/720 width matrix, canonical dark/light palettes, real avatars, true 44px controls, shared-prefix lane falsifiers, and native narrow disclosure. The operator selected the evolved D/synthesis instead, but the A/B+C artifacts remain useful design archaeology: they record what was evaluated, which traits were retained, and why the synthesis won.

Without a dedicated fold target, the PR remains permanently draft or must be closed without landing the evidence. Neither state matches the operator-directed draft drain.

The Architectural Reality

The owned surface is static design evidence under apps/agentos/design/; it does not modify AgentCard runtime code, Store/Model state, SCSS consumption, or visual-regression baselines.
PR #15538 / ticket #15561 own the selected D/synthesis design SSOT.
The future AgentCard/SCSS/CARD-CONTRACT/#14618 implementation carries the real Resolves #15536.
This fold preserves rejected alternatives and their falsifiers only. It must not reopen the selected product direction or become a second implementation authority.

Decision Record impact: none. This is a design-record fold under an already-settled operator selection.

The Fix

Reframe PR #15547 as the rejected-direction design record.
Add Resolves #<this ticket> while retaining Refs #15536.
Replace pre-selection residuals and post-merge gates with the settled operator-selection and evidence-record contract.
Flip PR #15547 ready, let exact-head CI rerun, and route Phoebe as the cross-family review seat.
After approval, hand merge eligibility to the human operator.

Acceptance Criteria

PR #15547 declares Resolves #<this ticket> and retains Refs #15536.
Its body names A and B/C as rejected-direction evidence and does not claim to resolve or supersede the selected D/synthesis.
PR #15547 is ready for review and exact-head CI is green.
A cross-family approval is recorded at the exact head.
Post-merge: both comparison artifacts are present on dev as design archaeology.

Out of Scope

Changing either comparison artifact after operator selection.
Screenshots; the operator explicitly made them optional.
The selected D/synthesis artifact, owned by PR #15538 / ticket #15561.
AgentCard runtime composition, SCSS, CARD-CONTRACT, Store/Model behavior, keyboard/touch behavior, or #14618 baselines.
Closing #15536; only the production implementation does that.

Avoided Traps

Using Resolves #15536 on an evidence PR — dishonest and incompatible with the one-resolves-per-ticket contract.
Treating rejected alternatives as a second design SSOT — the selected direction remains D/synthesis.
Blocking the fold on optional screenshots.
Reopening a fourth design cycle after the operator selected.

Parent design ticket: #15536
Rejected-direction PR: #15547
Selected-direction fold: #15561
Selected-direction PR: #15538
Superseded rejected-baseline control: #15539

Origin Session ID: a0518292-02c3-49ee-af08-adff40bc30b1

Retrieval Hint: AgentCard A B/C rejected directions fold bde617a320 operator selected D synthesis