graph-builder

@cognipeer/graph-builder · Utilities

stable

An API-first TypeScript library that turns text sources into a queryable graph. Build it from texts, providers, or change streams — and persist, update, or replay the result.

Quickstart GitHub Full docs

License · MITTypeScript · 100%Audience · Teams modelling document…Fits · Knowledge graphs

Knowledge graphsIncremental ingestionPath queriesBrowse recipes

Production install

$npm install @cognipeer/graph-builder

What's in the box

Teams modelling document relationships, knowledge graphs, or multi-step reasoning over a corpus that needs to update over time without re-ingesting everything. Each capability is opt-in — use the parts that fit, leave the rest.

API-first design

Embed it inside any app — no CLI required. The entire graph lifecycle is exposed as composable functions.

Many input modes

Build from inline texts, from a custom provider, or from a change stream that yields updates over time.

Incremental updates

`fromChanges` and `updateGraph` evolve an existing graph without re-processing the whole corpus.

Built-in analysis

Identify god nodes, run path queries, and inspect link clusters with the included query helpers.

Serializable

Load and persist graphs as plain JSON — easy to ship across services, store in a DB, or diff in a PR.

Adapters included

Out-of-the-box adapters for to-markdown sources and OpenAI-compatible providers; bring your own for anything else.

How it runs

Turn a corpus into a queryable graph in two ways: build from inline texts, or stream changes through a provider. Inspect, persist, and update incrementally.

Step 1

Source

texts · provider

Step 2

Build

graphBuilder()

Step 3

Analyse

god nodes · paths

Step 4

Persist

JSON snapshot

Step 5

Update

fromChanges()

Builders

fromTexts(records)
fromProvider(provider)
fromChanges(prev, provider)
updateGraph(prev, patch)

Adapters

to-markdown
OpenAI-compatible
Custom providers
Change streams

Query

analysis.godNodes
query.path(a, b)
artifacts.report
loadGraph(json)

Quickstart

Install, configure, run. The example below is the smallest piece of code that does something useful in production.

import  { graphBuilder } from "@cognipeer/graph-builder";
 
const result = await graphBuilder.fromTexts([
  { id: "docs/overview.md", title: "Overview",
    text: "# Overview&bsol;n&bsol;nSee [Guide](guide.md)." },
  { id: "docs/guide.md", title: "Guide",
    text: "# Guide&bsol;n&bsol;nThis explains the flow." },
]);
 
console.log(result.analysis.godNodes);
console.log(result.query.path("Overview", "Guide"));

How it compares

Against the utilities options teams most often weigh — focused on operational concerns, not feature inventories.

Capability	graph-builder	LangChain GraphIndexCreator	Neo4j + bespoke pipeline	Llama-Index KnowledgeGraph
Text → graph	● native	● native	● native	● native
Incremental updates	● native	○ missing	● native	◐ partial
Path queries	● native	◐ partial	● native	◐ partial
Provider adapters	● native	◐ partial	○ missing	◐ partial
TypeScript	● native	◐ partial	◐ partial	○ missing
Open source	● native	● native	◐ partial	● native