Copilot Superpowers

Browse, preview, and install 940+ AI agent skills for GitHub Copilot, Cursor, Windsurf and AntiGravity.

Install

1. Open VS Code Extensions (Ctrl+Shift+X)
2. Search "Copilot Superpowers" by aliabbaschadhar
3. Click Install — no config needed, works offline

Features

Install targets

Quick Start

• Press Ctrl+Shift+/ in VS Code
• Type a skill name (e.g. "react", "security", "aws")
• Press Enter — /<skill-id> is copied to your clipboard
• Paste into your editor's AI chat

Or open the sidebar, click any skill → Install to make it a persistent rule.

Skill Categories

ai-engineer · react-patterns · api-design-principles · aws-serverless · api-security-best-practices · android-jetpack-compose-expert · 3d-web-experience · sql-query-optimizer · kubernetes-patterns · and 930+ more.

Configuration

Search aiSkills. in VS Code Settings (Ctrl+,):

Editor Integration

After installing skills, add this to your editor in .github/copilot-instructions.md or .cursor/instructions.md or .windsurf/instructions.md or .antigravity/instructions.md so agent automatically discovers and uses them:

## AI Agent Skills

This project uses Copilot Superpowers skills stored in `.agent/skills/`.

### How to use skills

- Each installed skill lives at `.agent/skills/<skill-id>/SKILL.md`
- Before answering domain-specific questions, check if a relevant skill exists:
  - React / frontend → `.agent/skills/react-patterns/SKILL.md`
  - API design → `.agent/skills/api-design-principles/SKILL.md`
  - Security → `.agent/skills/api-security-best-practices/SKILL.md`
  - AI / agents → `.agent/skills/ai-agents-architect/SKILL.md`
  - …and so on for any other installed skill

### Loading a skill

Reference a skill file directly in your prompt:

```
#file:.agent/skills/react-patterns/SKILL.md
```

Or use the **Language Model Tools** for programmatic access:

| Tool                     | Description                           | Example                                                |
| ------------------------ | ------------------------------------- | ------------------------------------------------------ |
| `aiSkills_requestSkill`  | Install & load a skill's full content | `{"skillId": "react-patterns"}`                        |
| `aiSkills_listInstalled` | List all installed skills             | `{"category": "react", "limit": 10}`                   |
| `aiSkills_searchSkills`  | Search catalog by keyword             | `{"query": "aws", "limit": 15}`                        |
| `aiSkills_getSkillInfo`  | Get skill metadata                    | `{"skillId": "api-design", "includePreview": false}`   |
| `aiSkills_checkUpdates`  | Check for skill updates               | `{"skillId": "react-patterns"}` (or omit for all)      |
| `aiSkills_batchInstall`  | Install multiple skills               | `{"skillIds": ["react", "ts"], "category": "backend"}` |

Example tool call:

```
#tool:aiSkills_requestSkill {"skillId": "react-patterns"}
```

- If the skill is **already installed**, it loads instantly with no prompt.
- If the skill is **not installed**, VS Code will ask for confirmation before writing any file.

### Skill catalog

Browse all 940+ available skills:

- Open `assets/skills-catalog.md` (or `.agent/skills-catalog.md` after a sync)
- Or press `Ctrl+Shift+/` in VS Code to fuzzy-search and install skills

### Rules

1. Always check `.agent/skills/` before answering domain-specific questions.
2. When loading a skill, apply its guidance to your entire response.
3. Only reference skill IDs you can confirm exist in the catalog — never guess.

Tip: You can also auto-generate this block by running AI Skills: Copy Copilot Instructions from the Command Palette (Ctrl+Shift+P).

Language Model Tools (for AI Assistants)

This extension provides 7 Language Model Tools that GitHub Copilot and other AI assistants can use programmatically during chat sessions:

1. aiSkills_requestSkill — Install & Load Skill

Purpose: Install a skill (if needed) and load its full content into the AI's context.

Input Schema:

{
  "skillId": "string" // Required: The kebab-case skill ID
}

When to use: When you need domain-specific knowledge that a skill provides. Always check the catalog first before calling.

Example:

#tool:aiSkills_requestSkill {"skillId": "api-design-principles"}

2. aiSkills_listInstalled — List Installed Skills

Purpose: Discover what skills are already installed in the current workspace.

Input Schema:

{
  "category": "string", // Optional: Filter by category
  "limit": "number" // Optional: Max results (default: 100)
}

When to use: When you need to know what skills are available without installing new ones.

Example:

#tool:aiSkills_listInstalled {"category": "react", "limit": 20}

3. aiSkills_searchSkills — Search Skill Catalog

Purpose: Search the full catalog (940+ skills) by keyword using fuzzy matching.

Input Schema:

{
  "query": "string", // Required: Search keyword
  "category": "string", // Optional: Filter by category
  "limit": "number", // Optional: Max results (default: 20)
  "includeInstalled": "boolean" // Optional: Show install status (default: true)
}

When to use: When the user asks for skill recommendations or you need to discover relevant skills.

Example:

#tool:aiSkills_searchSkills {"query": "aws lambda", "limit": 10}

4. aiSkills_getSkillInfo — Get Skill Metadata

Purpose: Retrieve detailed information about a specific skill without installing it.

Input Schema:

{
  "skillId": "string", // Required: The skill ID
  "includePreview": "boolean" // Optional: Include content preview if installed (default: false)
}

When to use: When you need to verify a skill exists or get its description before installing.

Example:

#tool:aiSkills_getSkillInfo {"skillId": "react-patterns", "includePreview": false}

5. aiSkills_checkUpdates — Check for Updates

Purpose: Check if installed skills have newer versions available in the catalog.

Input Schema:

{
  "skillId": "string" // Optional: Check specific skill. Omit to check all installed skills.
}

When to use: Periodically or when the user asks about skill updates.

Example:

#tool:aiSkills_checkUpdates {}  // Check all installed skills
#tool:aiSkills_checkUpdates {"skillId": "react-patterns"}  // Check specific skill

6. aiSkills_batchInstall — Install Multiple Skills

Purpose: Install multiple skills in a single operation, either by ID list or by category.

Input Schema:

{
  "skillIds": ["string"], // Optional: Array of skill IDs to install
  "category": "string", // Optional: Install all skills in a category (overrides skillIds)
  "overwrite": "boolean" // Optional: Overwrite existing skills (default: false)
}

When to use: When setting up a project with multiple skills or installing an entire category.

Example:

#tool:aiSkills_batchInstall {"skillIds": ["react-patterns", "typescript-best-practices"]}
#tool:aiSkills_batchInstall {"category": "api-design"}  // Install all API design skills

7. aiSkills_planInstall — Plan Install from Intent

Purpose: Convert a natural-language request into recommended skill IDs and optionally install them in one step.

Input Schema:

{
  "query": "string", // Required: Natural language request
  "limit": "number", // Optional: Max recommendations (default: 8)
  "includeInstalled": "boolean", // Optional: Include already installed skills (default: true)
  "install": "boolean", // Optional: Install recommendations after confirmation (default: false)
  "overwrite": "boolean", // Optional: Overwrite existing installed files (default: false)
  "outputFormat": "\"markdown\"|\"json\"" // Optional: Response format (default: "markdown")
}

When to use: When the user asks to install a set of skills by intent (e.g. "react testing", "API security").

Example:

#tool:aiSkills_planInstall {"query": "react testing", "install": false}
#tool:aiSkills_planInstall {"query": "secure REST APIs", "install": true, "outputFormat": "json"}

Tool Usage Best Practices

1. Check catalog first — Only call tools with skill IDs you can confirm exist
2. No arbitrary limits — You can call multiple tools but ensure they're relevant
3. Call early — Invoke tools at the start of your response so content is in context
4. Handle errors gracefully — If a tool fails, proceed without retrying
5. Respect user confirmation — Installation tools show confirmation dialogs for uninstalled skills
6. Use JSON when chaining tools — Set outputFormat: "json" for machine-readable responses

Contributing & Support

• Bugs / features → GitHub Issues
• Pull requests → see CONTRIBUTING.md
• Security → see SECURITY.md
• Changelog → CHANGELOG.md

MIT License · Built by aliabbaschadhar ❤️