starlake-ai

Starlake Skills — Data skill for Claude Code

Data community

Skills for Starlake data pipeline CLI.

How to install Starlake Skills

This entry records only its repository, not the path inside it, so there is no exact command to give. Open starlake-ai/starlake-skills and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Starlake Skills does

Skills for Starlake data pipeline CLI. Covers load, transform, extract, autoload, dag-generate, configuration patterns, connection types, write strategies, data quality, and best practices for DuckDB, BigQuery, Snowflake, Airflow & Dagster. 🚧 Work in progress.

Alternatives in Data

Ars Full — ARS full pipeline — research → write → review → revise → finalize 4.5k ★
Skill Content Pipeline — Extract patterns and anatomy from URLs — use to reverse-engineer content strategies from live pages 2.8k ★
Check Data Quality — Renamed → /portaljs-check-data-quality 2.3k ★

README

Starlake Skills

**Agent Skills bundle** for the [Starlake](https://starlake.ai) data pipeline CLI: a core skill for every CLI command and configuration pattern, plus the optional Starflow guided methodology with workflow skills and expert personas for data engineering.

Covers all CLI commands (load, transform, extract, autoload, dag-generate...), configuration patterns, connection types, write strategies, data quality expectations, and production best practices for DuckDB, BigQuery, Snowflake, Redshift, and Airflow/Dagster orchestration.

Installation

There are two ways to install, and both end in the same state: every skill folder in this bundle is **symlinked** into the skills directory of each AI assistant, where it is discovered automatically.

Release install (end users): downloads a versioned archive from GitHub Releases into ~/.starlake-skills, then links the skills. No git or Node required, only curl and tar (PowerShell 5.1+ on Windows).
Git clone (contributors): links point straight into your clone, so edits to any SKILL.md and every git pull are live immediately, with no re-install.

What gets installed where

Link created	Points to	Purpose
`~/.claude/skills/`	`.agents/skills/` and `.agents/starflow/skills/`	Skill discovery for Claude Code
`~/.copilot/skills/`	same	Skill discovery for GitHub Copilot
`~/.gemini/skills/`	same	Skill discovery for Gemini CLI
`~/./starflow`	`.agents/starflow`	Starflow config, templates, and the config-resolver scripts

Each assistant reads the `name` and `description` frontmatter of every linked `SKILL.md` to decide when a skill applies; no build step or registration is involved. Because links are directories, step files, templates, and configs inside a skill folder come along automatically. The installer only ever removes symlinks that resolve back into this bundle: it ne