Skip to content

HelioAI

AI agent for heliophysics and space plasma data analysis.

Ask a question in natural language. HelioAI finds the right parameter across 70+ missions, downloads it, runs the analysis in a sandbox, and hands back a plot, a number, and a notebook you can re-run.

You:      "IP shock in WIND data, January 2005 — compute θ_Bn"

HelioAI:  → resolves parameter IDs for B, Vp, Np across 83k speasy products
          → downloads the time series via speasy (AMDA / CDAWeb / CSA)
          → runs shock detection + the coplanarity theorem in a sandboxed Python env
          → returns a plot, the θ_Bn value, and a reproducible .ipynb notebook

No API key is needed for data access, and no manual parameter hunting.

Why it exists

Finding the right parameter is often harder than the analysis itself. A researcher who knows exactly what they want — the magnetic field magnitude from Cluster 3 during this substorm — still has to know that it lives under c3_b, in which dataset, from which provider, in which coordinate system. HelioAI closes that gap with a hybrid semantic + lexical search over the whole speasy catalogue, then carries on into the analysis instead of stopping at the download.

What makes it different

  • Automation-first. The unit of work is a scientific question, not a plot panel. Point it at an event catalog and it will run the same analysis across every event.
  • Provenance by default. Derived quantities come from recipes that each carry a citation, and every session exports a Methods & data acknowledgements section listing what was used.
  • Reproducible output. Any session becomes a self-contained .ipynb whose cells run in a plain Jupyter kernel, with sandbox helpers rewritten to direct speasy calls.
  • No separate AI subscription. Bring any provider — Azure OpenAI, Groq, Gemini, or a fully local Ollama model.

Where it sits in the ecosystem

HelioAI is a layer on top of the existing Python heliophysics stack, not a replacement for any of it:

It uses For
speasy data access across AMDA, CDAWeb and CSA
PlasmaPy plasma physics formulary
geopack coordinate transforms and boundary models
Astropy · SunPy units, time, solar context

Catalogs are written in the standard speasy format, so an event set detected here opens directly in SciQLop for visual inspection.

Next steps

  • Install it — pip install helioai-agent, then build the parameter index once.
  • Quickstart — your first question, end to end.
  • Interfaces — CLI, Jupyter, web UI, or MCP server.

License and citation

MIT. If HelioAI contributes to published work, please cite it — see CITATION.cff — and cite the underlying data providers and any recipe references the export lists for you.