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	13819
title	Embed ADR nodes for first-class hybrid-GraphRAG queryability (ADR 0024 §2.8 / 0023 §2.5 target)
state	Closed
labels	enhancementaiarchitecturemodel-experience
assignees	neo-opus-grace
createdAt	Jun 21, 2026, 10:42 PM
updatedAt	Jun 21, 2026, 11:47 PM
githubUrl	https://github.com/neomjs/neo/issues/13819
author	neo-opus-grace
commentsCount	0
parentIssue	null
subIssues	[]
subIssuesCompleted	0
subIssuesTotal	0
contentTrust
projected
quarantined	0
signals	[]
blockedBy	[]
blocking	[]
closedAt	Jun 21, 2026, 11:47 PM

Embed ADR nodes for first-class hybrid-GraphRAG queryability (ADR 0024 §2.8 / 0023 §2.5 target)

Closed v13.1.0/archive-v13-1-0-chunk-5 enhancementaiarchitecturemodel-experience

neo-opus-grace commented on Jun 21, 2026, 10:42 PM

Context

The ADR-node embedding target named in BOTH ADR 0024 §2.8 (#13815) AND ADR 0023 §2.5 (#13806). ADR 0006 / #11377 (CLOSED) made ADRs graph-queryable entities (AdrIngestor inserts ADR nodes); this is the un-done follow-up — making them semantically queryable (embedded → first-class in hybrid GraphRAG).

The Problem (V-B-A'd at dev)

AdrIngestor inserts every ADR as a first-class ADR node (with GOVERNS/CITES_AUTHORITY/IMPLEMENTS_DECISION/GRADUATED_FROM/CODIFIES_CONCEPT edges) but does not set semanticVectorId — git grep -E "semanticVectorId|embed" origin/dev -- ai/services/ingestion/AdrIngestor.mjs returns nothing. So the live graph's 23 ADR nodes are un-embedded → inert to the semantic candidate pool (query_hybrid_graph / search_nodes). An agent asking "what governs the golden path?" cannot surface ADR 0023/0024 via semantic query — the architecture is not self-documenting in the graph.

The Fix

AdrIngestor embeds each ADR's text (title + decision/summary) via the same embedder the REM extractor uses, sets semanticVectorId, and writes the vector to Chroma — so ADR nodes surface as first-class architectural anchors in hybrid GraphRAG. Dogfooding: the graph indexes its own decision records.

Acceptance Criteria

AdrIngestor sets semanticVectorId on each ADR node (embed title + decision text).
The vector is written to the Chroma store (queryable via query_hybrid_graph / search_nodes).
An ADR-governance query (e.g. "golden path map-fidelity") surfaces ADR 0023/0024.
Unit test: a stubbed ADR → embedded → semanticVectorId set + vector written.
Re-embedding is idempotent (re-ingest does not duplicate vectors).

Out of Scope

The canonical edge-type registry (ADR 0024 §6 follow-up). The content-as-SSOT unification (0024 §2.6). Any node/edge taxonomy change.

Aligned-with: ADR 0024 §2.8 (#13815, the embed-to-first-class target) + ADR 0023 §2.5 (#13806, roadmapped). Builds on: ADR 0006 / #11377 (CLOSED — the insertion). Model ADR: #13814.

Live latest-open sweep: gh issue search "ADR node embed" → no equivalent open issue at 2026-06-21T20:40.

Origin Session ID: 80932414-00dc-4d2e-96e6-d0bcf1529733

Retrieval Hint: query_raw_memories("ADR node embedding semanticVectorId AdrIngestor first-class hybrid-GraphRAG queryable dogfooding 0024 2.8")

tobiu referenced in commit ad73fdd - "feat(ai): embed ADR nodes into the graph collection for hybrid-GraphRAG queryability (#13819) (#13820) on Jun 21, 2026, 11:47 PM

tobiu closed this issue on Jun 21, 2026, 11:47 PM