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.

Neo.mjs - Self-Evolving Software Organism

Frontmatter

title	>-
author	tobiu
state	Merged
createdAt	Apr 15, 2026, 12:55 PM
updatedAt	Apr 15, 2026, 1:07 PM
closedAt	Apr 15, 2026, 1:07 PM
mergedAt	Apr 15, 2026, 1:07 PM
branches	`dev` ← `agent/10023-tech-debt-radar`
url	https://github.com/neomjs/neo/pull/10026

Merged

tobiu commented on Apr 15, 2026, 12:55 PM

This PR formalizes the Tech Debt Radar, systematically mapping and targeting architectural debt using frontier-model-driven semantic RAG sweeps across historical issues and Agent OS memory sessions. It also updates the ticket intake and PR review gates to mandate 'Semantic Blast-Radius Sweeps' prior to architectural changes, protecting against unknown unknowns. Resolves #10023

tobiu commented on Apr 15, 2026, 1:00 PM

Input from Gemini 3.1 Pro (Antigravity):

✦ # PR Review Summary

Status: Approved

Excellent work putting this together. The implementation perfectly targets the "unknown unknowns" problem by weaponizing semantic RAG sweeps at critical architectural junctures. This is exactly the kind of self-correcting capability the Agent OS needs.

📊 Evaluation Metrics

[ARCH_ALIGNMENT]: 100 - Superb alignment. Adheres strictly to the Progressive Disclosure pattern for skills and correctly injects constraints into existing workflow gates.

[CONTENT_COMPLETENESS]: 95 - The guide is extremely comprehensive, correctly specifying Model Tiering (Opus/Pro) and detailing multiple vector sweeps. The updates to CodebaseOverview and ProgressiveDisclosureSkills successfully tied the loop.

[EXECUTION_QUALITY]: 95 - The markdown is precise, well-structured, and explicitly outlines the multi-vector sweep execution strategy.

[PRODUCTIVITY]: 100 - Achieves all objectives of #10023, effectively formalizing the Tech Debt Radar.

[IMPACT]: 90 - Very high. This introduces a structural paradigm shift from reactive to proactive debt-scanning.

[COMPLEXITY]: 75 - Involved spanning multiple meta-documents, updating established reflection gates, and drafting a comprehensive execution stricture.

[EFFORT_PROFILE]: Architectural Pillar - This skill structurally shifts how swarm agents operate and establishes systemic debt remediation loops.

🕸️ Context & Graph Linking

Target Epic / Issue ID: Resolves #10023

Related Graph Nodes: Tech Debt Radar, ticket-intake, pr-review.

🧠 Graph Ingestion Notes

[KB_GAP]: None observed directly in the initial execution. However, the subsequent "missing skills inventory" fix identified a temporary gap in operational context logic (forgetting to update upstream indexes when creating new domain entities), mapping an important meta-learning point.

[TOOLING_GAP]: N/A

[RETROSPECTIVE]: The proactive RAG blast-radius checking instituted here is highly robust. Explicitly gating architectural features (refactor(ai)) behind an enforced historical context lookup is a powerful mitigation against AI amnesia and tunnel vision. Model Tiering ensures the cognitive complexity of this task is only handled by capable Frontier Models.

📋 Required Actions

To proceed with merging, please address the following:

Integrate standard Progressive Disclosure indexing into CodebaseOverview.md and ProgressiveDisclosureSkills.md (Already resolved via follow-up commits).

Everything looks fully squared away. LGTM!