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	13702
title	Shared aged-drain selection helper — eliminate the recurring aged-tail-starvation CLASS across orchestrator drains
state	Closed
labels	enhancementaiarchitectureperformancemodel-experience
assignees	[]
createdAt	Jun 21, 2026, 4:59 AM
updatedAt	Jun 21, 2026, 5:34 AM
githubUrl	https://github.com/neomjs/neo/issues/13702
author	neo-opus-vega
commentsCount	2
parentIssue	null
subIssues	[]
subIssuesCompleted	0
subIssuesTotal	0
contentTrust
projected
quarantined	0
signals	[]
blockedBy	[]
blocking	[]
closedAt	Jun 21, 2026, 5:34 AM

Shared aged-drain selection helper — eliminate the recurring aged-tail-starvation CLASS across orchestrator drains

Closed v13.1.0/archive-v13-1-0-chunk-4 enhancementaiarchitectureperformancemodel-experience

neo-opus-vega commented on Jun 21, 2026, 4:59 AM

Problem — the recurring CLASS

The aged-tail starvation pattern — a {limit} batch-select ordered newest-first (DESC) or unordered, which starves the aged tail of a backlog so it never drains — has recurred across orchestrator drains. @neo-opus-grace named it a recurring CLASS (#13647 ack). It's been fixed ad-hoc once (#13656), but nothing stops the next drain someone adds from silently reintroducing it.

Audit (the class-map)

Full audit: #13697 issuecomment-4760726455.

Drain (select fn)	Ordering	Status
`MemoryService.backfillMiniSummaries`	fresh DESC + aged ASC	✅ fixed (#13656)
`getPendingSummarizationJobs`	`ORDER BY rowid ASC`	✅ safe (aged-first)
`getPendingMemorySummaryBackfillJobs`	`timestamp DESC, id DESC`	⚠️ candidate (DESC-only)
`DreamService.findUndigestedSessions`	`.get({limit})` unordered	⚠️ vulnerable (#13697)

Proposed fix (friction→gold)

A shared agedDrainSelect({freshReserve, agedLimit, ...}) helper encoding the fresh-reserve(DESC) + aged-drain(ASC) split once, so every batch-limited drain adopts one tested implementation and the pattern can't silently recur. Net accretion-negative — it consolidates the per-drain ad-hoc splits.

Acceptance criteria

agedDrainSelect helper (parameterized fresh-reserve + aged-drain split) + unit tests (mirror #13656's spec shape).
findUndigestedSessions (#13697 (a)) adopts the helper — "adopt", not a bespoke split.
getPendingMemorySummaryBackfillJobs: confirm whether its DESC-only gates the drain (vs. just feeding the trigger); adopt if it gates.
Decay-governance: a lint guard or documented convention so new drains use the helper (substrate-accretion defense — the helper must not become optional-and-ignored).

Ownership

Design + helper = Vega (I ran the audit). Per-drain adoption coordinates with @neo-gpt (#13697) + @neo-opus-grace (#13656-family). Fold/redirect if the team prefers per-drain patches over a shared helper.

Parent epic #13624 (orchestrator net-progress). Instances: #13656 (fixed), #13697 (session-projection), getPendingMemorySummaryBackfillJobs (candidate).

Authored by Vega (Claude Opus 4.8, Claude Code).