A Python notebook that lives in one HTML file.

KERNEL is a complete notebook — cells, real Python, plots, DataFrames, a variable inspector, data mounting, .ipynb round-trip — built on Pyodide and packed into a single self-contained .html. No install, no build step, no server.

Open the file and you have a kernel. Nothing leaves your machine — the Python runs in your tab, your data stays in the page.

Launch KERNEL → Launch the Agent → View source

Pyodide downloads ~10 MB of WebAssembly on first run (cached after). Desktop browser recommended.

Two ways to use it

notebook · agent

KERNEL Live

The notebook itself. Write Python and markdown, run cells in execution order, get plots, interactive Plotly, rendered DataFrames, KaTeX math and Mermaid diagrams. Mount CSVs and files, inspect every variable, export to .ipynb — all client-side.

Launch KERNEL →

No account, no key, nothing to configure.

KERNEL Agent Live

A natural-language, human-in-the-loop analyst that builds the notebook for you — writing cells, running them, and seeing the results (text and figures) to iterate. Bring your own Anthropic API key; it's stored in your browser and sent only to Anthropic.

Launch the Agent →

Bring your own key · stored locally · read the spec

What's in the box

capabilities

Real PythonPyodide / CPython 3.x with numpy, pandas, matplotlib, scipy, scikit-learn, sympy, statsmodels and more.
Ordered outputstdout, figures, rich displays and results stream back in true execution order.
Rich markdownGFM tables, fenced code, KaTeX math ( $…$ ) and Mermaid diagrams.
Display helpersdisplay, display_html, display_plotly for interactive output.
Data workspacedrop in CSV / JSON / files, preview them, insert a read snippet in a click.
Variable inspectorlive namespace with types, sizes, and click-to-expand detail (head, dtypes, stats).
Composable panelsnotebooks, data, variables and split outputs dock, resize, and collapse to a calm center.
Portableround-trips to .ipynb, exports to .py, persists locally — one file you can host anywhere.

The kernel-notebooks skill

for Claude

A Claude skill that teaches the model to author exceptional notebooks for this runtime — well-structured, narrated analyses that use exactly the capabilities KERNEL has and none it doesn't. It's also the system prompt behind the agentic build: how to run the loop, read text and image outputs back, and explore with a human in the loop.

Claude.ai: download kernel-notebooks.skill, then Settings → Capabilities → Skills → upload.
Claude Code: copy the skill/ folder into your skills directory:

git clone https://github.com/dogum/kernel.git
cp -r kernel/skill ~/.claude/skills/kernel-notebooks