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	13895
title	[msg-ingestion] Drain crash-recovery replay + idempotency + retention + observability
state	Closed
labels	enhancementaiarchitecturemodel-experience
assignees	[]
createdAt	Jun 23, 2026, 1:17 AM
updatedAt	Jun 23, 2026, 1:22 AM
githubUrl	https://github.com/neomjs/neo/issues/13895
author	neo-opus-grace
commentsCount	1
parentIssue	null
subIssues	[]
subIssuesCompleted	0
subIssuesTotal	0
contentTrust
projected
quarantined	0
signals	[]
blockedBy	[x] 13893 [msg-ingestion] Durable message WAL + drain daemon (local + cloud parity)
blocking	[]
closedAt	Jun 23, 2026, 1:22 AM

[msg-ingestion] Drain crash-recovery replay + idempotency + retention + observability

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

neo-opus-grace commented on Jun 23, 2026, 1:17 AM

Parent: Epic #13889. The robustness layer that makes the durability invariant hold under failure.

Problem / scope

A durable WAL is only as good as its drain's failure-handling. The daemon must survive crashes mid-drain (replay un-drained entries), not double-ingest (idempotency), bound the WAL (retention past the drain watermark), and be observable — a silent stalled drain is exactly what hid de1d9f0b.

Acceptance Criteria

Replay: on restart, the daemon re-drains WAL entries not yet confirmed-persisted (a crash mid-drain loses nothing).
Idempotency: re-ingesting a WAL entry is a no-op (the message node is upserted by stable id, never duplicated).
Retention: the message WAL is pruned past the confirmed-drain watermark (no unbounded growth), mirroring the memory-wal reconciliation/prune.
Observability: drain lag / pending-count / failures are observable (a recordTaskOutcome-style trace or healthcheck field) — so a stalled drain is loud, not silent.

Contract Ledger

Adds: replay + watermark + retention + drain observability on the message-ingestion daemon.
Reuses: the embed-daemon reconciliation/prune patterns (pruneReconciledWalSegments).

Sequencing

Blocked-by the WAL + drain daemon sub (extends the daemon). Can land after the add_message flip.