Amplify Active Hybrid GraphRAG in v13 notes

Context

The v13 release notes were improved in #12794, but the Dream / Native Edge Graph section still under-explains the active part of Active Hybrid GraphRAG.

The current draft correctly says that episodic memory becomes semantic entities, codebase structure becomes graph topology, and the current frontier becomes an advisory forecast. That is accurate, but it compresses the most distinctive part: GraphService is not only a graph lookup service. It exposes an active frontier API (getContextFrontier, mutateFrontier) that lets agents read, pivot, decay, and hand off the current strategic context through topology.

The operator also pointed out that the release notes can quote the generated Golden Path snapshot directly. That matters because resources/content/sandman_handoff.md is concrete proof that the Dream / Sandman pipeline is not just architecture prose: it emits an auditable handoff, including its own gap warnings.

Live latest-open sweep: checked latest 20 open issues on 2026-06-09 via gh issue list --state open --limit 20 --json number,title,author,labels,url; no equivalent narrow release-note follow-up found. Related parent: #12696. Existing adjacent tickets include #12065 (REM / Sandman pipeline), #12073 (Tri-Vector hierarchy), and #12456 (AiConfig SSOT), but none targets the v13 release-note narrative.

Semantic KB sweep: attempted ask_knowledge_base("v13 release notes Active Hybrid GraphRAG Golden Path Sandman handoff mutate_frontier GraphService narrative"), but the Knowledge Base was unavailable (Failed to access collections). Local content sweep covered resources/content/issues, resources/content/discussions, resources/content/release-notes, and learn/agentos for Active Hybrid GraphRAG, Golden Path, sandman_handoff, mutate_frontier, mutateFrontier, and Topological Context.

The Problem

The release notes currently give Dream / Golden Path a credible section, but the strongest mechanism is still too implicit:

• Active Hybrid GraphRAG is named in the market/moat paragraph, but not unpacked as a mechanism.
• mutate_frontier / mutateFrontier and getContextFrontier are not visible enough as active graph primitives.
• AiConfig using Neo.state.Provider, the graph using Neo.data.Store / Neo.data.Model, and vector routing through class-based service boundaries are not tied clearly enough to the Brain/Body substrate story.
• resources/content/sandman_handoff.md is mentioned as generated proof, but the release notes do not yet quote the Golden Path snapshot, which would make the proof feel alive.
• The Dream section has one Mermaid diagram, but the release notes can use one or two more diagrams where they clarify the architecture rather than becoming an architecture slide dump.

The Architectural Reality

Verified source anchors:

• ai/services/memory-core/GraphService.mjs — getContextFrontier() reads immediate high-weight frontier neighbors, filters closed/RLS-invisible paths, and returns the active context frontier.
• ai/services/memory-core/GraphService.mjs — mutateFrontier() anchors a new strategic target from frontier, decays older outbound frontier paths by 50%, and writes a new weighted edge.
• ai/graph/Database.mjs — the Native Edge Graph uses Neo.data.Store for node/edge caches, with NodeModel / EdgeModel and indexed source/target edge lookup.
• learn/agentos/AiConfigModel.md — Neo.ai.ConfigProvider extends Neo.state.Provider; the Brain config is a provider hierarchy using the same state-provider ownership model as the Body.
• ai/services/graph/GoldenPathSynthesizer.mjs — Golden Path synthesis combines a frontier baseline vector, Chroma semantic candidates, SQLite Native Edge Graph topology, structural scoring, and handoff rendering.
• resources/content/sandman_handoff.md — generated proof surface for Dream / Golden Path output and graph self-audit warnings.

The Fix

Update resources/content/release-notes/v13.0.0.md only.

Prescribe these narrative changes:

1. Add a compact subsection inside ## The Dream: The Graph Started Forecasting explaining Active Hybrid GraphRAG: Recall Became Steering.
2. Explain that Chroma answers semantic proximity while the Native Edge Graph answers structural pressure, and that the active frontier bridges those two layers.
3. Name getContextFrontier() and mutateFrontier() as the active primitives, without making them the main v13 headline.
4. Add a Mermaid diagram for the Active Hybrid GraphRAG loop.
5. Quote the current Golden Path snapshot from resources/content/sandman_handoff.md as generated proof, while preserving the caveat that Golden Path is advisory and peer-reviewed, not a central scheduler.
6. Add one concise sentence tying the Brain’s implementation back to Neo’s own primitives: Neo.data.Store, Neo.data.Model, Neo.state.Provider, and class-based service boundaries.

Acceptance Criteria

• resources/content/release-notes/v13.0.0.md includes an Active Hybrid GraphRAG subsection inside the Dream section.
• The subsection explains the active frontier without claiming this is the main v13 headline.
• The text explicitly names getContextFrontier() and mutateFrontier() / mutate_frontier as frontier primitives.
• The Dream section includes a Mermaid diagram for the Active Hybrid GraphRAG loop.
• The release notes quote a current Golden Path / sandman_handoff.md snapshot as generated proof.
• The Brain/Body substrate point is present but compact: Agent OS uses Neo’s own data, state, and class primitives.
• The section avoids overclaims such as “zero sync overhead,” “death of state machines,” “scale flawlessly,” or “the graph knows what to do.”
• No unrelated release-note sections or PR-count numbers are churned.

Out of Scope

• No changes to GraphService, GoldenPathSynthesizer, DreamService, AiConfig, or MCP tools.
• No attempt to regenerate sandman_handoff.md.
• No broad rewrite of the v13 release-note spine.
• No new benchmark claims for Gemma4 unless a measured number already exists in committed docs and is cited precisely.
• No PR-body narrative polish beyond the standard required template.

Avoided Traps

• Do not make Topological Context the headline of v13; it supports the Brain/Dream story.
• Do not frame v13 as anti-state. The point is that the Agent OS validates Neo’s state/data/class substrate.
• Do not turn Golden Path into a command hierarchy. It is a graph-backed forecast that peers choose, challenge, and review.
• Do not over-diagram. Mermaid diagrams should explain mechanisms, not decorate the prose.

Related

• Parent release epic: #12696
• Related REM / Sandman epic: #12065
• Related Tri-Vector work: #12073
• Related AiConfig SSOT epic: #12456
• Prior release-note PR: #12794

Origin Session ID: 720fc09a-e8ff-467d-b039-28120312964d

Handoff Retrieval Hint: v13 release notes Active Hybrid GraphRAG mutate_frontier Golden Path sandman_handoff Dream section