@tekmemo/graph

The graph memory engine for TekMemo. This package provides the logic for managing entities (nodes) and their relationships (edges).

Install

npm install @tekmemo/graph

Core Concepts

• Nodes: Represent discrete entities like a Decision, File, or Requirement.
• Edges: Represent the relationships between nodes, such as depends_on, implements, or supersedes.
• JSONL Storage: Nodes and edges are stored as JSON Lines for easy appending and stream-parsing.

API Reference

Store Helpers

Helper	Purpose
createInMemoryGraphStore(options?)	Initializes a fast, in-memory graph store.
parseGraphNodesJsonl(content)	Parses raw JSONL text into GraphNode objects.
parseGraphEdgesJsonl(content)	Parses raw JSONL text into GraphEdge objects.

Graph Operations

Helper	Purpose
expandFromEntities(store, ids, options)	Traverses the graph to find a neighborhood of related nodes.
extractGraphFactsRuleBased(input)	Extracts nodes and edges from text using configured rules.

Example: Managing a Graph

import { createInMemoryGraphStore } from "@tekmemo/graph";

const store = createInMemoryGraphStore();

// Add a node
await store.nodes.create({
  id: "node_1",
  type: "decision",
  label: "Use D1 Database",
  metadata: { status: "accepted" }
});

// Add another node
await store.nodes.create({
  id: "node_2",
  type: "decision",
  label: "Use Tailwind CSS"
});

// Create an edge between them
await store.edges.create({
  source: "node_1",
  target: "node_2",
  type: "relates_to"
});

// Query neighbors
const neighbors = await store.neighbors({ id: "node_1" });
console.log(`Found ${neighbors.nodes.length} connected entities.`);

Use Cases

• Impact Analysis: See what parts of the system are affected by a change.
• Architectural Discovery: Help agents understand how different decisions relate to each other.
• Context Grounding: Provide a "neighborhood" of relevant facts to an LLM instead of a flat list.