Projects / AI Tools

Artifactor

How do I turn my notes into a proper document?

A TypeScript library that turns conversations, notes and raw text into structured documents using an LLM and a template you define. It is aimed at things like resumes, forms and reports.

ForDevelopers building apps that need to produce structured documents from messy input.
SolvesGetting a language model to fill a specific document structure, keep it valid, and update it over time takes a lot of glue code.

What it does

You describe a document as a template: a JSON Schema-like structure, instructions for the model (a role and a list of guidelines), a completion check and optional quality metrics. Artifactor turns that template into an agent. The agent creates an artifact instance, and you feed it unstructured text such as a paragraph of someone describing their career. The instance fills in the structured data, reports how complete it is and suggests what is missing.

How it works

An artifact instance is a live document. You can update it with more text or an editing instruction like “make this more professional”, set a field directly, or read a part of it back. Every change is kept in a version history with rollback.

Instances emit events, so other code can react to changes. Artifacts can also subscribe to each other: a cover letter can watch the experience and skills fields of a resume and update itself automatically when they change.

Providers

The library is LLM-agnostic. It takes any OpenAI-compatible client, with examples for OpenAI, Anthropic and local models. It runs in Node.js 16 or later and in the browser through a bundler.

Status

Published to npm as artifactor under the MIT license. The repo has not been updated since September 2025.