JSONL Schema & Concept Seed Extraction

Objective

Define the JSONL schema for concept nodes and edges, and populate the initial concept graph by extracting from existing curated sources.

Context

Phase 1, Step 1 of the Concept Ontology epic (#10030). The concept graph starts as standalone JSONL files — no database dependency.

Deliverables

1. JSONL Schema Definition

Create nodes.jsonl and edges.jsonl at .neo-ai-data/concepts/:

Node schema (one concept per line):

{"id":"multi-threading","name":"Multi-Threading Architecture","tier":1,"description":"...","uniqueToNeo":true,"tags":["architecture","workers"]}

Edge schema (one relationship per line):

{"source":"two-tier-reactivity","target":"push-reactivity","type":"PARENT_CONCEPT"}
{"source":"push-reactivity","target":"file-src/core/Config.mjs","type":"IMPLEMENTED_BY"}

Edge types: PARENT_CONCEPT, IMPLEMENTED_BY, EXPLAINED_BY, EXEMPLIFIED_BY, REQUIRES

2. Gitignore Exception

Add negation rule to .gitignore:

!.neo-ai-data/concepts/

3. Concept Extraction Sources

• Primary: learn/tree.json (filtered — skip DevIndex, Tutorials, JavaScript basics, Comparisons)
• Secondary: learn/benefits/ArchitectureOverview.md headings → Tier 1 concepts
• Tertiary: Human curation pass to validate Teaching Test filter

4. README.md

Schema documentation and contribution guide at .neo-ai-data/concepts/README.md.

The Teaching Test (Filter)

A concept is included if:

1. A developer needs to understand it to USE Neo.mjs productively
2. It cannot be learned by simply reading one API doc page
3. It answers "how" or "why" questions, not just "what"

Expected Output

~80-100 CONCEPT nodes with hierarchy, ~200 PARENT_CONCEPT edges, concept descriptions from existing docs.

Acceptance Criteria

• nodes.jsonl and edges.jsonl exist and are valid JSONL
• .gitignore negation rule works (files tracked by git)
• Concepts pass the Teaching Test
• README documents the schema

Origin Session ID: 496c0120-29aa-4b02-a804-d5a32ff6b5eb