uberSKILLS

Design, test, and deploy Agent Skills through a visual, AI-assisted workflow.

uberSKILLS replaces manual SKILL.md authoring with a structured editor, multi-model testing sandbox, and one-click deployment to your agent's skills directory.

What is uberSKILLS?

Creating Agent Skills today is entirely manual -- authors hand-write YAML frontmatter and markdown instructions, with no built-in way to preview, validate, or test a skill before deploying it.

uberSKILLS is an open-source web application that provides an integrated authoring environment purpose-built for the SKILL.md format, with deploy support for Antigravity, Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, OpenCode, and Windsurf:

• Skills Library -- Browse, search, filter, and manage all your skills in one place
• AI-Assisted Creation -- Describe what you want in natural language; get a complete SKILL.md draft
• Structured Editor -- Edit metadata, instructions, and files with real-time validation and auto-save
• Multi-Model Testing -- Test skills against any model on OpenRouter with streaming responses and metrics
• Export & Deploy -- One-click deploy to 8 agent tools including Claude Code, GitHub Copilot, Cursor, and Windsurf, or export as zip
• Import -- Bulk-import existing skills from directories or zip archives
• Version History -- Every change is versioned; browse and compare past versions

Tech Stack

Quickstart

Run uberSKILLS with a single command — no cloning, no setup:

npx @uberskillsdev/uberskills

That's it. On first run it will automatically set up everything (~2 minutes), then start the server at http://localhost:3000.

Your data is stored at ~/.uberskills/data/ and persists across runs.

Options

--port <number>    Port to run on (default: 3000)
--host <string>    Host to bind to (default: localhost)
--data-dir <path>  Custom data directory (default: ~/.uberskills/data/)
--reset            Delete cached install and re-setup
-d, --debug        Enable debug log level

Prerequisites

• Node.js >= 20
• git — available in PATH

Development Setup

If you want to contribute or run from source:

# Clone the repository
git clone https://github.com/uberskillsdev/uberskills.git
cd uberskills

# Install dependencies (pnpm is enabled via corepack)
corepack enable
pnpm install

# Start the development server
pnpm dev

Open http://localhost:3000. On first launch the app will automatically:

1. Create the data/ directory and SQLite database
2. Run Drizzle migrations to initialize the schema
3. Generate an encryption secret at data/.secret
4. Open the Settings page prompting for an OpenRouter API key

No Docker, no Postgres, no signup required.

Project Structure

uberskills/
├── apps/
│   ├── web/                 # Next.js 15 — Skills editor app (port 3000)
│   └── landing/             # Next.js 15 — Static landing page (port 3001)
├── packages/
│   ├── types/               # Shared TypeScript interfaces
│   ├── db/                  # Database schema, client, queries (Drizzle + SQLite)
│   ├── skill-engine/        # SKILL.md parser, validator, generator, importer, exporter
│   └── ui/                  # Shared UI components, ThemeProvider, design tokens (shadcn/ui)
├── docs/                    # Project documentation
├── turbo.json               # Turborepo pipeline configuration
├── biome.json               # Linter and formatter configuration
└── tsconfig.json            # Base TypeScript configuration

Available Scripts

Environment Variables

Set these in apps/web/.env.local or apps/landing/.env.local (not committed to version control).

Detailed Documentation

Full project documentation is available in the docs/ directory:

• Getting Started -- Installation, setup, and first steps
• Features Guide -- How to use the skills library, editor, testing, import/export
• Architecture -- Monorepo structure, package responsibilities, data flow
• API Reference -- REST API endpoints and response formats
• Database -- Schema, Drizzle ORM patterns, queries
• AI Integration -- OpenRouter, Vercel AI SDK, streaming
• Skill Engine -- Parser, validator, generator, importer, exporter
• Security -- Encryption, input validation, filesystem safety
• Deployment -- Docker, Vercel, self-hosting options

Docker

Build and run via Docker:

docker compose up -d

The app will be available at http://localhost:3000. The data/ directory is mounted as a volume for persistence.

See Dockerfile and docker-compose.yml for details.

Contributing

Contributions are welcome! Please read the Contributing Guide for details on our development workflow, coding standards, and how to submit pull requests.

License

MIT