Skip to content

Welcome

lightcone-cli is a small toolchain that turns a research question into a reproducible analysis. You describe what you're trying to learn, a cli agent helps you turn that into a precise specification, and the lc command line keeps the resulting code, decisions, and outputs in sync — forever.

You don't need to write Python or YAML by hand. The agent handles the implementation; you stay in charge of the scientific choices.

What this guide covers

  • Install — get lc and Claude Code running on your machine.
  • Getting Started — create your first project, run it end-to-end, and understand what each piece does.
  • The Agentic Workflow — /lc-new, /lc-from-code, /lc-from-paper, and /lc-feedback — what each one does and when to reach for it.
  • Running on a Cluster — taking your analysis to a SLURM HPC system, including Perlmutter-specific notes.
  • Troubleshooting — common issues and how to unstick them.
  • Glossary — the terms that show up everywhere (universe, decision, manifest, …) explained in plain language.

What you'll do, in three lines

Quick start

uv tool install lightcone-cli
lc init my-analysis && cd my-analysis
claude                                 # then, inside Claude Code: /lc-new

pip install lightcone-cli
lc init my-analysis && cd my-analysis
claude                                # then, inside Claude Code: /lc-new

That's the shortest possible path. The rest of the guide is the unhurried version.

What lightcone-cli is not

  • A statistics package. It runs your code; it doesn't compute things itself.
  • A workflow language. Recipes in astra.yaml are short shell or Python commands, not a DSL. There's no learning curve beyond what's in Getting Started.
  • An IDE. lc is a command-line tool; the agent surface lives inside the agent harness (Claude Code for now).

If you'd rather skim the design and architecture, the maintainer docs are the other half of this site.