Skills
A curated set of skills for the work data scientists and researchers repeat every week: research with sources, exploring data, spreadsheets, statistics, and figures. Written for Claude Code, Codex, Cursor, and Runcell inside Jupyter.
Turn a question into a multi-step research run with sources, cross-checks and a cited report.
Runcell teamFree
VisualizationChart choice, consistent styling and journal-ready export for matplotlib, seaborn and plotly.
Runcell teamFreeFrom question and data to the right test, with assumptions checked and results reported correctly.
Runcell teamFreeEach page describes what the skill does, what it produces, and the checks it runs along the way.
Turn a question into a multi-step research run with sources, cross-checks and a cited report.
Runcell teamFreeRead, clean and analyze spreadsheets with pandas. Reproducible answers, not a rewritten sheet.
Runcell teamFreeLiterature search, citation verification and structured notes for agents that write with references.
Runcell teamFreeVisualizationChart choice, consistent styling and journal-ready export for matplotlib, seaborn and plotly.
Runcell teamFreeFrom question and data to the right test, with assumptions checked and results reported correctly.
Runcell teamFreeA systematic first pass over any dataset: structure, quality, distributions, relationships, and what to do next.
Runcell teamFreeA skill lives in your agent's skills folder. When a task matches its description, the agent reads the SKILL.md and follows the procedure inside.
Steps, checks and output formats are spelled out, so a research run or a statistical test is done the same way each time, with its reasoning visible.
In Runcell the agent executes in a live Jupyter kernel. Tables, figures and reports come from cells you can inspect and rerun.
A skill is a folder with a SKILL.md file that describes a procedure an AI agent should follow for a specific kind of task, plus any scripts or reference files it needs. Agents such as Claude Code, Codex and Cursor load the skill when the task matches, so the same job is done the same way every time.
Yes. They use the Agent Skills format that Claude Code, Codex, Cursor and Runcell all read. What differs between agents is the tools available to them, for example web search or a Python kernel, and each skill states what it needs.
Large collections cover hundreds of libraries and domains. This set is small and focused on the workflows data scientists and researchers repeat every week: exploring a dataset, choosing a statistical test, making figures, doing research with citations. Each skill is written to run inside a Jupyter notebook as well as in a terminal agent.
Most of these workflows produce data, tables and figures, not just text. In a notebook the agent executes each step in a live kernel, so results are visible, checkable and rerunnable. Runcell is a Jupyter-native agent built for exactly that.
Runcell is a Jupyter-native AI agent. Every step of a skill runs in a live kernel, so the result is a cell you can rerun.