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	10172
title	Graph node-ID case canonicalization: settle memory:/MEMORY: and session:/SESSION: convention
state	Closed
labels	enhancementaiarchitecturecore
assignees	neo-gpt
createdAt	Apr 22, 2026, 1:19 AM
updatedAt	3:18 PM
githubUrl	https://github.com/neomjs/neo/issues/10172
author	tobiu
commentsCount	3
parentIssue	10143
subIssues	[]
subIssuesCompleted	0
subIssuesTotal	0
contentTrust
projected
quarantined	0
signals	[]
blockedBy	[]
blocking	[]
closedAt	Jun 6, 2026, 1:23 AM

Graph node-ID case canonicalization: settle memory:/MEMORY: and session:/SESSION: convention

Closed v13.0.0/archive-v13-0-0-chunk-5 enhancementaiarchitecturecore

tobiu commented on Apr 22, 2026, 1:19 AM

Context

The Native Edge Graph previously carried two case conventions for row-backed Memory and Session node IDs:

Ingestor/storage convention: lowercase memory:<chromaId> and session:<sessionId>.
Extractor/provenance prompt convention: uppercase MEMORY:<chromaId> and SESSION:<sessionId>.

Current source and tests have already converged most runtime behavior toward lowercase storage. The remaining work is contract/documentation alignment, not a graph-wide prefix migration.

Current State (2026-06-03 V-B-A)

MemorySessionIngestor writes row-backed Memory/Session graph IDs as lowercase memory:<id> and session:<id>.
GraphService.normalizeGraphNodeId() lowercases only Memory/Session prefixes and leaves other prefixes such as CONCEPT: / CLASS: / AGENT: outside this ticket.
Current tests assert uppercase Memory/Session inputs normalize to lowercase storage/linking rather than persisting uppercase duplicates.
learn/agentos/tooling/GraphBackfill.md still needs the canonical-format section updated from deferred/ambiguous wording.
SemanticGraphExtractor provenance/lazy-queue examples and predicates still need reconciliation with the row-backed lowercase Memory/Session contract.

Decision

Use Option A for this ticket: canonical row-backed Memory/Session storage IDs are lowercase memory:<id> and session:<id>. Uppercase MEMORY:<id> / SESSION:<id> remain compatibility input forms only.

This ticket explicitly does not migrate semantic or identity prefixes such as CONCEPT:, CLASS:, AGENT:, or other non-row-backed node families.

Contract Ledger Matrix

Target Surface	Source of Authority	Proposed Behavior	Fallback / Edge Case	Docs	Evidence
Row-backed Memory node IDs	Current `MemorySessionIngestor` + #10172 decision	Canonical storage ID is `memory:<chromaId>`.	Uppercase `MEMORY:<id>` accepted as compatibility input and normalized before storage/linking.	`GraphBackfill.md` canonical format section	Unit/source verification that storage remains lowercase
Row-backed Session node IDs	Current `MemorySessionIngestor` + #10172 decision	Canonical storage ID is `session:<sessionId>`.	Uppercase `SESSION:<id>` accepted as compatibility input and normalized before storage/linking.	`GraphBackfill.md` canonical format section	Unit/source verification that storage remains lowercase
`GraphService.normalizeGraphNodeId()`	Existing compatibility shim	Retain as backward-compatible input normalization for Memory/Session only.	Unknown or non-row-backed prefixes pass through unchanged unless a separate graph-wide prefix policy ticket exists.	JSDoc if wording is stale	Focused tests for Memory/Session normalization and non-Memory/Session passthrough
`SemanticGraphExtractor` provenance / lazy-queue producer	Current extractor prompt/predicate + #10172 decision	Reconcile Memory/Session examples and provenance target detection with lowercase canonical row-backed IDs while preserving compatibility with uppercase queued inputs.	Existing uppercase queued entries still drain through normalization.	JSDoc/prompt text if changed	Fixture/unit coverage for queued provenance targets
`GraphBackfill.md` operator guidance	Operator/manual-query surface	Document lowercase Memory/Session canonical IDs and uppercase compatibility inputs so manual SQLite/graph queries do not miss rows.	Mention semantic uppercase prefixes are separate and out of scope.	`learn/agentos/tooling/GraphBackfill.md`	Docs grep + review evidence
Explicit exclusions	#10172 scope boundary	Do not migrate `CONCEPT:`, `CLASS:`, `AGENT:`, or other non-row-backed prefixes.	If future evidence needs graph-wide prefix settlement, file a separate ticket.	Ticket body / docs note	No source changes touching those prefixes under this ticket

Acceptance Criteria

Canonical decision recorded: lowercase memory: / session: for row-backed Memory/Session storage.
GraphBackfill.md documents canonical lowercase Memory/Session IDs and uppercase compatibility input forms.
SemanticGraphExtractor provenance/lazy-queue Memory/Session examples or predicates are reconciled with the canonical contract, if current source still emits misleading uppercase row-backed targets.
GraphService.normalizeGraphNodeId() remains as a backward-compatible shim and is documented/tested as Memory/Session-only.
No migration or recasing is applied to CONCEPT:, CLASS:, AGENT:, or other non-row-backed prefixes.

Out of Scope

Graph-wide prefix policy.
Agent identity prefix convention.
DB migration from lowercase to uppercase.
Recasting semantic node IDs such as CONCEPT: / CLASS:.

Predecessor shipped work: #10151, #10152, #10153, PR #10171.
Parent: #10143.
Grand-parent: #9999.
Review origin: Gemini 3.1 Pro PR review on #10171.

Origin Session ID: d69ac7a0-9fe8-4416-b766-cd9edb8bee71.

tobiu added the enhancement label on Apr 22, 2026, 1:19 AM

tobiu added the ai label on Apr 22, 2026, 1:19 AM

tobiu added the architecture label on Apr 22, 2026, 1:19 AM

tobiu added the core label on Apr 22, 2026, 1:19 AM

tobiu added parent issue #10143 on Apr 22, 2026, 1:19 AM

tobiu cross-referenced by PR #10171 on Apr 22, 2026, 1:20 AM

tobiu cross-referenced by #9999 on Apr 22, 2026, 11:09 AM

tobiu referenced in commit 07655f1 - "fix(memory): canonicalize memory session graph ids (#10172) (#12601)" on Jun 6, 2026, 1:23 AM

tobiu closed this issue on Jun 6, 2026, 1:23 AM