Skip to main content

Documentation Index

Fetch the complete documentation index at: https://hubify.com/docs/llms.txt

Use this file to discover all available pages before exploring further.

​
Welcome to Hubify Labs

Hubify Labs is a Scientific Discovery Platform that gives researchers AI-powered infrastructure to run experiments, manage multi-agent teams, write papers, and publish discoveries, all from a single IDE-like interface.

​
What is Hubify Labs?

Think of it as Cursor for research. Instead of writing code, you’re running experiments, analyzing data, and publishing papers, with AI agents doing the heavy lifting.

Captain View

Your research command center. See experiments, tasks, agents, and papers at a glance.

Multi-Agent System

Orchestrator + lead agents + workers. Cross-model peer review. Automatic task routing.

GPU Compute

RunPod H100/H200 integration. Auto-scaling pods. Cost optimization.

Paper Pipeline

From experiment to arXiv-ready PDF. LaTeX compilation, figure generation, peer review.

​
Available on Every Surface

Hubify Labs runs everywhere:
  • Web UI, Full-featured research IDE at hubify.com
  • Desktop App, Native macOS app (Windows coming soon)
  • CLI, hubify command for terminal-first researchers
All three surfaces are equivalent, same features, same data, same agents. Pick the one that fits your workflow.

​
Quick Example

# Install the CLI
npm install -g @hubify/cli

# Login
hubify auth login

# Create a lab
hubify lab create --name "my-research" --template cosmology

# Run an experiment
hubify experiment run --name "mcmc-chain-1" --pod h100

# Check status
hubify status

​
Key Features

FeatureDescription
LabsIsolated research environments with their own agents, data, and compute
ExperimentsGPU-powered experiment execution with automatic logging and QC
AgentsMulti-agent system with orchestrator, leads, and workers
PapersEnd-to-end paper pipeline from data to arXiv submission
Knowledge BaseKarpathy-style structured wiki that grows with your research
Lab SitesPublic research website at .hubify.app
Novelty ScoringAI-powered assessment of how novel your findings are

​
Getting Started

1

Create an account

Sign up at hubify.com/sign-up
2

Create your first lab

Follow the quickstart guide to set up a research lab
3

Configure agents

Set up your agent team with the right models and roles
4

Run experiments

Launch your first experiment on GPU compute

​
Architecture & Internals

For developers, contributors, and agent builders working against Hubify Labs internals, the canonical architecture references live in the hubify repo docs/:
  • architecture.md, six-layer source-of-truth model: GitHub (lab filesystem), Hubify DB (structured state), Gateway (async coordination), RunPod (compute), MCP/CLI/API/SDK (agent access), Lab Sites (always-visible truth layer)
  • github-sync.md, round-trip flow: managed repo creation, user-owned migration, webhook in, orchestrator commits out
  • event-types.md, durable agentEvents ledger: full type catalog, idempotency rules, mirror set, recipe for adding a new type
  • runpod-compute.md, dispatch flow with budget gates, approval, recommendation, pod lifecycle, cost events
  • lab-sites.md, Lab Site as the always-visible truth layer; publish-mode taxonomy; subdomain + custom-domain roadmap
  • paper-pipeline.md, 14-stage paper pipeline from claim map to published
  • adversarial-peer-review.md, multi-role cross-model review protocol (skeptic, methods, narrative, novelty, reader)
  • agents.md, orchestrator, lead, worker roles + the local-agent workflow (Claude Code, Codex, Cursor, OpenClaw)
  • local-agent-workflow.md, how Claude Code / Codex / gh CLI consume the lab repo and round-trip back into Hubify
  • deployment-tiers.md, cloud / hybrid / sovereign deployment direction
  • open-core-strategy.md, open-core split rationale + module boundaries
  • huggingface-roadmap.md, HuggingFace publish integration roadmap
  • lab-repo-manifest.md, hubify-lab/v1 manifest schema (hubify.json)
The starter lab repo template (README.md, hubify.json, lab.md, agent.md, CLAUDE.md, full project-context/ set, placeholder dirs) lives in lab-scaffold/. The matching agent skill catalog is at skills/hubify-labs/.