Reusable AI agent skills for practical data, analytics and architecture delivery.
PowerData Skills is a public library of structured Markdown skills designed to help AI assistants, coding agents and data professionals apply repeatable delivery methods across data engineering, analytics engineering, data modelling, lakehouse architecture and operational data use cases.
These skills are not just prompts. They are lightweight delivery procedures that define when to use a skill, what inputs are required, what decisions need to be made, and what a useful output should look like.
- Data engineers
- Analytics engineers
- BI developers
- Solution and data architects
- AI-assisted development practitioners
- Teams wanting repeatable AI workflows for data projects
An AI agent skill is a structured, reusable instruction set for an AI assistant or coding agent. It defines:
- When to use it — the context in which the skill applies
- What inputs are needed — the information to provide before starting
- What the AI should do — a clear operating procedure, not a vague instruction
- What the output should look like — a consistent, predictable structure
Skills are written in Markdown and are designed to be copied into AI tools, referenced as instructions, or loaded as context in agentic workflows.
| Skill | Description |
|---|---|
| Use Case to Data Requirements | Translate a business use case into structured data requirements |
| Dimensional Model Designer | Design star schema and dimensional models for BI and reporting |
| Medallion Architecture Designer | Design bronze/silver/gold lakehouse data layers |
| Data Pipeline Designer | Design high-level data pipelines for source-to-target data flows |
| Skill | Description |
|---|---|
| Flet + Supabase App Framework | Scaffold a Flet + Supabase multi-platform Python app with correct project structure and integration patterns |
| Flet Multi-Platform Build | Configure Android, iOS, and web Docker build pipelines for a Flet app with CI/CD |
| Flet ACA Deploy | Deploy a Flet web app to Azure Container Apps |
| App Icon Asset Generation | Generate a consistent application icon asset set from an approved high-resolution logo |
| Skill | Description |
|---|---|
| Time-Series Use Case Assessment | Assess time-series use cases and translate them into data requirements |
skills/
README.md
LICENSE
CONTRIBUTING.md
.gitignore
docs/
skill-authoring-guide.md # How to write a good skill
usage-patterns.md # How to use skills across AI tools
databricks-genie-code.md # Using skills with Databricks Genie Code
roadmap.md # Planned additions
templates/
skill-template/ # Blank template for new skills
skills/
data/ # Foundation skills for data delivery
app/ # Skills for multi-platform app development
domain/ # Domain-specific skill packs
Skills are plain Markdown files. There is no installation required. You load them into your AI tool by pasting or referencing the skill content.
Skillfish is a CLI tool that installs skills into every AI coding agent on your machine in one command:
npx skillfish add POWR-DATA/skillsThis works with Claude Code, Cursor, Windsurf, GitHub Copilot, Gemini CLI, and 30+ other agents.
Claude.ai (web): Paste this into a new conversation:
I have a set of reusable skills here:
https://github.com/POWR-DATA/skills
Please read the skills, summarise what is available, and use them
automatically when relevant in this conversation. Confirm once loaded.
Claude Code: Copy a skill into your project and reference it directly:
mkdir -p .claude/skills
cp -r /path/to/skills/skills/data/dimensional-model-designer .claude/skills/Then in Claude Code: "Apply the dimensional model designer skill to this use case."
Any other AI tool: Open a SKILL.md file, copy the content, and paste it as instructions into your tool of choice.
For full installation and usage guidance across all supported tools, see the Documentation section below.
| Document | Description |
|---|---|
| docs/usage-patterns.md | How to use skills across Claude, Cursor, Skillfish, GitHub Copilot, and other tools |
| docs/databricks-genie-code.md | Installing and using skills with Databricks Genie Code |
| docs/skill-authoring-guide.md | How to write a well-structured skill |
| docs/roadmap.md | Planned additions to the library |
Contributions are welcome. See CONTRIBUTING.md for guidance.
This repository is intentionally generic and public-safe. Skills represent generalised patterns for common data delivery tasks.
Organisation-specific implementations should be created privately by extending these baseline skills with internal standards, platform constraints, naming conventions, security requirements, and governance requirements.
PowerData Skills is maintained by PowerData.
For more context about the skills library, visit the AI Agent Skills Library.
MIT. See LICENSE.