Skip to main content

​
Research OS Template

Research OS is available as a template and ready for deployment. Select it during workspace creation at hubify.com/signup. Research OS is built for researchers, academics, and anyone doing deep knowledge work. The agent thinks in terms of papers, citations, synthesis, and knowledge graphs. Where other templates focus on tasks and communications, Research OS focuses on understanding and building knowledge over time.

​
What’s Included

​
Skills

SkillWhat It Does
arxivMonitors arXiv for new papers matching your research interests, downloads and indexes PDFs
perplexityDeep web research — uses Perplexity AI for comprehensive source gathering
knowledge-hubManages curated knowledge collections in the knowledge/ directory
synthesisCombines multiple sources into structured synthesis documents with citations

​
Dashboard Focus

SectionWhat It Shows
PapersPaper queue — new, reading, read, synthesized — with search and filters
SynthesisGenerated synthesis documents combining multiple papers on a topic
NotesResearch notes organized by topic and date
Knowledge GraphVisual map of concepts, papers, and their relationships

​
Research Pipeline

Research OS structures knowledge work into a four-stage pipeline:
1

Discover

Your agent monitors arXiv, academic databases, and web sources for papers matching your research interests. New papers are added to your queue automatically with a relevance score.
2

Read

Papers in your queue are pre-processed — key sections highlighted, abstract summarized, methodology extracted. You can read the full paper or consume the agent’s summary.
3

Synthesize

When you have read multiple papers on a topic, your agent generates synthesis documents — structured analyses that compare methodologies, identify consensus and contradictions, and surface gaps.
4

Publish

Synthesis documents and research notes are stored in your knowledge hub. Optionally contribute insights to the global intelligence layer for other researchers.

​
The Knowledge Hub

The knowledge hub is the core data structure of Research OS. It lives in the knowledge/ directory and contains curated, evergreen knowledge — not dated logs, but structured understanding that improves over time. 
knowledge/
  INDEX.md                    # Auto-generated topic index
  transformer-architectures.md
  attention-mechanisms.md
  scaling-laws.md
  rlhf-techniques.md
  multimodal-fusion.md
Each knowledge file follows a consistent structure: 
# Topic Name

## Summary
One-paragraph overview of the topic.

## Key Papers
- [Paper Title](link) — one-line summary of contribution
- [Paper Title](link) — one-line summary of contribution

## Current Understanding
What the field currently believes, with confidence levels.

## Open Questions
What is not yet resolved or understood.

## Your Notes
Personal observations and hypotheses.

## Last Updated
Auto-updated when new papers or synthesis touch this topic.
The knowledge hub is fundamentally different from the memory/ directory. Memory stores time-based episodic logs (what happened when). Knowledge stores evergreen understanding (what is true, what matters). Both are searchable, but they serve different purposes.

​
Model Routing

Research OS uses models optimized for deep reading and synthesis: 
models:
  default: "anthropic/claude-sonnet-4-6"

  routing:
    orchestrator: "anthropic/claude-sonnet-4-6"   # Coordination and planning
    researcher: "kimi/k2.5"                         # Fast, large context — good for paper ingestion
    synthesizer: "anthropic/claude-sonnet-4-6"     # Quality synthesis and writing
    automation: "google/gemini-flash"               # arXiv monitoring, notifications
The researcher role uses a large-context model for ingesting full papers, while the synthesizer uses a high-quality model for producing accurate, well-structured synthesis documents.

​
When to Use Research OS

Choose Research OS if you:
  • Read academic papers regularly and want them organized
  • Need synthesis across multiple sources on the same topic
  • Build and maintain a personal knowledge base
  • Want your agent to monitor arXiv or academic databases for you
  • Do deep research as a core part of your work
Choose something else if you:
  • Primarily write code — use Dev OS
  • Do GTM and content work — use Founder OS
  • Want a general-purpose personal OS — use MyOS

​
SOUL.md Personality

# SOUL.md — Research OS

You are a research-focused AI operating system for {{USERNAME}}.

## Core Traits
- Precision over speed — accuracy matters more than quick answers
- Citation-first — always reference sources, never make unsupported claims
- Synthesis-oriented — connect ideas across papers and disciplines
- Intellectually curious — flag interesting tangents and related work

## Priorities
1. Monitor paper sources and maintain a quality reading queue
2. Pre-process papers for efficient human review
3. Generate synthesis documents when sufficient material accumulates
4. Maintain the knowledge hub with current understanding
5. Surface contradictions and open questions in the literature

## Communication Style
- Dashboard for paper queue and synthesis review
- Knowledge hub for evergreen understanding
- Telegram for high-relevance paper alerts only

​
Next Steps

All Templates

Compare all 6 templates side by side

AI OS Overview

How workspaces and the reserved structure work