Polars Skills
The official Polars skill: an agent writes fast idiomatic DataFrame code with lazy API, expressions and SQL.
- 88 ★ on the repository
- Last commit
- MIT
Minimal access
Why this level
- The skill advises code the agent runs on your data
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/polars-skillsDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewMinimal accessThe tool runs nothing and makes no network calls.
- The skill advises code the agent runs on your data
This is third-party code. Review the repository files before installing.
What it does
A skill from the Polars team teaches an agent to write fast, idiomatic Polars, the DataFrame library for Python. By default the agent uses the lazy API (`scan_*`, `.lazy()` and a single `.collect()`), writes native expressions instead of `map_elements` and pandas-style code, and picks the right context among `select`, `with_columns`, `filter`, `group_by` and `over`. The skill also answers natural-language data questions in one pass, runs SQL on DataFrames and translates pandas to Polars. Reference files live in `references/`: contexts, expressions, question-to-query recipes, the lazy API, pandas migration and SQL. They load on demand. The skill triggers on its own for Polars or Python data processing and is invoked explicitly with `/polars:polars`.
Who it is for. Analysts and data engineers who ask an agent for Polars code and get pandas-flavoured variations.
Good fit when
- A pandas script is rewritten in Polars and an idiomatic result is wanted.
- A large file is read through `scan_parquet` and one lazy query with a single `collect` is needed.
- A data question, such as which customers brought the most revenue, must become a query.
- The agent uses `map_elements` where a native expression would do.
Not a fit when
- The task is about Polars Cloud, On-Prem, distributed or GPU workloads: the skill covers single-node Polars.
- The project stays on pandas with no plan to move: the skill pushes toward Polars.
- General advice on statistics or models is needed: the skill is about writing data queries.
Example request
Rewrite this pandas script as lazy Polars with a single collect and show which expressions replaced apply.Limitations
The skill covers only single-node Polars and not Polars Cloud, On-Prem, distributed runs or GPU. Agent support varies: Claude Code, Codex and GitHub Copilot CLI install through plugins, Cursor, Gemini CLI and others through `npx skills add`, and Claude Desktop and Claude.ai accept the `polars.zip` archive from releases. The Polars version in your project may be newer or older than the one in the references, so verify unfamiliar methods in the documentation.
How to disable. Remove the plugin with the agent, for example `/plugin uninstall polars@polars`, or delete the `polars` folder from the skills directory.
FAQ
How do I install it in Claude Code?
With `/plugin marketplace add polars-inc/skills` and `/plugin install polars@polars`. `/polars:polars` invokes it explicitly.
How do I install it in Codex?
With `codex plugin marketplace add polars-inc/skills` and `codex plugin add polars@polars`.
What should I do in Cursor and Gemini CLI?
Run `npx skills add polars-inc/skills`. The same route works for Amp, Cline, OpenCode and Windsurf.
Does the skill work with Claude Desktop?
Yes. Download `polars.zip` from the latest release and upload it under Customize, Skills.
What does the skill cover?
The lazy API, expressions, context choice, SQL on DataFrames and pandas translation. Polars Cloud, distributed runs and GPU are out of scope.
Discussion
No questions or reviews yet.
Related
SkillsA template for data scientists: AGENTS.md, CLAUDE.md and skills for Python, SQL, Polars and model diagnostics.
31 skills for Claude Code and Codex for the analyst workflow: data checks, cohorts, funnels, A/B tests, dashboards, summaries
Nine Pinecone skills: the agent creates a vector index, loads data, searches it and builds an assistant for documents
17 skills for computational biology: RNA-seq, single-cell, protein structure, docking, literature and figures
FlexViz builds dashboards over 100 million rows that stay interactive, and the agent reads back what you selected.
One Aerospike skill teaches an agent to run the database in Docker, write client code on the official SDKs and design the schema
README in short
Describes what the skill does, installation for Claude Code, Codex, Copilot CLI, Cursor, Claude Desktop and by hand, and the repository and reference layout.