A collection of reusable AI agent skills, instructions, and shared knowledge for GitHub Copilot and Claude. Each skill packages domain expertise, best practices, and step-by-step workflows so AI assistants produce consistent, high-quality output across projects.
ai-skills/
├── skills/ # Reusable AI skills, organized by domain
│ ├── backend/
│ │ ├── dotnet/ # code-review, generate-crud, refactor-clean-architecture, write-unit-test
│ │ └── node/
│ ├── frontend/
│ │ ├── angular/
│ │ └── nextjs/ # generate-api-route, generate-page, refactor-component, sso-with-google
│ ├── cross/ # Language-agnostic skills
│ │ ├── code-review/
│ │ ├── create-user-story/
│ │ ├── debug/
│ │ └── explain-code/
│ └── workflow/ # Multi-step process skills
│ ├── build-feature/
│ ├── fix-bug/
│ └── full-code-review/
├── agents/ # Agent configurations
│ ├── claude/ # Claude system prompts and templates
│ └── copilot/ # GitHub Copilot agent configs
├── shared/ # Reference knowledge consumed by skills
│ ├── architecture/ # clean-architecture, microservices, monolith
│ ├── best-practices/ # error-handling, logging, validation
│ └── coding-standards/ # angular, dotnet, general, nextjs
├── schemas/ # JSON/YAML schemas for skill validation
├── sdk/ # Tooling for authoring and testing skills
└── examples/ # Example prompts and usage demos
| Skill | Trigger | Description |
|---|---|---|
create-user-story |
Writing a Jira user story ticket | Creates a well-structured Jira user story following Product Team best practices — user perspective, value hypothesis, acceptance criteria |
code-review |
Reviewing code quality | — |
debug |
Diagnosing a bug | — |
explain-code |
Explaining what code does | — |
| Skill | Description |
|---|---|
code-review |
.NET-specific code review |
generate-crud |
Generate CRUD endpoints for a resource |
refactor-clean-architecture |
Refactor code to follow Clean Architecture layers |
write-unit-test |
Write unit tests for a class or method |
| Skill | Trigger | Description |
|---|---|---|
sso-with-google |
Google SSO / NextAuth | Implement, fix, or review Google SSO using NextAuth.js v4 |
generate-api-route |
Creating a Next.js API route | — |
generate-page |
Creating a Next.js page | — |
refactor-component |
Refactoring a React component | — |
| Skill | Description |
|---|---|
build-feature |
End-to-end workflow for building a new feature |
fix-bug |
Structured workflow for diagnosing and fixing a bug |
full-code-review |
Multi-pass comprehensive code review |
Each skill lives in its own folder as a SKILL.md file with YAML frontmatter:
---
name: skill-name
description: "When to use this skill. USE FOR: ... DO NOT USE FOR: ..."
---
# Skill Title
[Step-by-step instructions for the AI agent]The description field is the discovery surface — the agent reads it to decide whether to load the skill. Use clear trigger phrases in USE FOR.
Copy the skill folder to .github/skills/<skill-name>/SKILL.md in your project, or reference it from your copilot-instructions.md.
Use skills as system prompt additions or include them in the agents/claude/ configuration.
- Create a folder under the appropriate domain:
skills/<domain>/<skill-name>/ - Add a
SKILL.mdwith YAML frontmatter (name,description) and step-by-step instructions - Reference shared knowledge from
shared/where applicable - Register the skill in
index.yaml
-
namematches the folder name -
descriptionincludes clearUSE FORandDO NOT USE FORtriggers - Instructions are written as directives to the AI, not to the human
- Includes a concrete example or output template
- No project-specific secrets or hardcoded paths
The shared/ folder contains reference documents that skills can import for consistent standards:
architecture/— Clean Architecture, Microservices, Monolith patternsbest-practices/— Error handling, Logging, Input validationcoding-standards/— Language and framework conventions (Angular, .NET, Next.js, General)