Skip to content

Scriptorium

Overview

Scriptorium is an open-source academic platform that transforms the traditional research library from a passive storage box into an active research partner. Think of it as "NotebookLM on steroids" built specifically as a daily driver for PhD students and researchers.

Unlike current tools that require you to manually search for papers and sync them across disjointed apps, Scriptorium provides a dedicated workspace — a modern scriptorium. It proactively monitors global research databases (OpenAlex, Semantic Scholar), curates reading lists based on your specific semantic footprint, and provides an augmented Markdown IDE where every generated sentence is deterministically linked back to a raw source highlight.

The Three Pillars

Pillar A: Continuous Discovery (The FOMO Killer) — A background "Watchdog" system continuously queries open academic graphs using the semantic profile of your current project. It alerts you when new, highly relevant papers are published or when your core citations are referenced by new work.

Pillar B: Granular Knowledge Graph (Highlights as Objects) — When you highlight a paragraph in the integrated PDF reader, it is instantly converted into a discrete database object containing the raw text, its vector embedding, and a relationship edge back to the source paper.

Pillar C: The Augmented IDE (Context-Aware Drafting) — A collaborative Markdown editor that maintains a sliding window of your recent text, injecting CitationHighlight objects into the LLM context for real-time, deterministic auto-completions. Hovering over any generated claim reveals the original PDF snippet for instant verification.

Status

RefAIn is in the Vision & Research phase, with design and experimentation happening in the associated GitHub repository.

  • Skill Networks — SkillNet's knowledge graph representation informs Scriptorium's literature management pipeline.