Skip to content

About

DomoSkills is a developer-native discovery engine and CLI installer for AI Agent Skills. Discover modular capabilities, curate your project's skill stack, and install directly into your AI coding assistant workspace in a single command.

Topics

Resources

Stars

12 stars

Watchers

0 watching

Forks

Latest commit

 

History

56 Commits

Folders and files

Repository files navigation

DomoSkills — The Open Agent Skills Marketplace

DOMOSKILLS_
Open skills. Smarter agents. Your stack.
DomoSkills App Icon
DomoSkills Animated Mascot

The free, open-source capability registry and CLI package manager for AI Agent Skills.

License: MIT Node Version Turborepo TypeScript Next.js 14 PRs Welcome

DomoSkills is a developer-native discovery engine and CLI installer for AI Agent Skills. Discover modular capabilities, curate your project's skill stack, and install directly into your AI coding assistant workspace in a single command.

Whether you build with Google Antigravity, Claude Code, Cursor IDE, OpenAI Codex/Agents, GitHub Copilot, or OpenCode Interpreter, DomoSkills standardizes capability discovery, frontmatter validation, AST security scanning, and reproducible lockfile management across your entire team.


Table of Contents


Quick Start

1. Discover Skills

Explore the open web registry at http://localhost:3000/explore or search from your terminal:

npx domoskills search react
npx domoskills search security

2. Add Skills to Your Project

Install single or multiple skills:

# Universal standard (default)
npx domoskills add react-performance owasp-agent-guardian

# Target a specific AI coding agent
npx domoskills add nextjs-app-router --agent claude
npx domoskills add fastapi-clean-architecture --agent cursor
npx domoskills add rag-pipeline-architect --agent opencode

3. Initialize & Diagnose Workspace

npx domoskills init
npx domoskills doctor
npx domoskills audit
npx domoskills list

Key Features

  • ⚡ 1,000+ Production-Ready Skills: Spanning 12 engineering categories—Frontend, UI/UX/Design, Backend, Fullstack, Security, DevOps, Cloud, AI/ML, Database, Testing, Mobile, and Developer Productivity.
  • 🔌 Universal Multi-Agent Adapter Engine: Native zero-configuration target directory mappings for 7 major AI environments: Universal Standard, Anthropic Claude Code, OpenAI Codex, Cursor IDE, OpenCode Interpreter, GitHub Copilot, and Google Antigravity / Gemini CLI.
  • 🛡️ Zero-Execution AST Security Engine: Static pattern analyzer checks every skill against OWASP guidelines, detecting destructive commands (rm -rf /), shell pipes (curl | bash), reverse shells, encoded PowerShell, and exposed secret keys.
  • 🔒 Path Traversal Protection: Hardened path normalization preventing ../ directory escapes and absolute path injections.
  • 📦 Reproducible Skill Lockfile (domoskills.json): Commit your team's AI capability dependencies to version control. Teammates run npx domoskills install to get identical agent behaviors.
  • 🧰 Full-Featured Developer CLI: 9 commands (init, search, add, install, remove, list, doctor, audit, update) built with Commander and chalk for fast, scriptable workflows.
  • 🌐 Modern Web Discovery Portal (apps/web): Next.js 14 App Router web platform featuring:
    • Instant Search & Multi-Faceted Filtering: Filter by category, target agent, security tier, and sort by popularity, rating, or score.
    • Density View Modes: Toggle seamlessly between Card Grid, Compact Data Table, and List views.
    • Side-by-Side Skill Comparator: Compare up to 4 skills simultaneously on file size, security score, author, license, and trigger words.
    • Interactive Skill Stack Cart: Add skills to a temporary drawer with a fly-to-cart micro-interaction, curating custom compound installation commands.
    • Global Command Palette (Cmd+K / Ctrl+K): Instant keyboard navigation across registry catalogs and platform tools.
    • Interactive Terminal Simulator: Live web terminal demonstrating CLI operations with copyable commands and realistic shell responses.
    • Web Agent Doctor (/doctor): Interactive browser health check scanning installed agent directories and environment variables.
    • Community Submission Studio (/submit): Visual SKILL.md authoring studio with real-time YAML frontmatter linting and AST security pre-flight checks.
    • Sound Effects & UI Motion: Optional tactile Web Audio feedback on actions, animated Domo mascot, and wavy marquee displays.

Architecture

DomoSkills is organized as a high-performance Turborepo monorepo:

domoskills/
├── apps/
│   └── web/                   # Next.js 14 App Router web platform & registry UI
├── packages/
│   ├── cli/                   # Executable `domoskills` CLI installer
│   ├── registry/              # Search engine, seed database, and ingestion providers
│   ├── skill-parser/          # SKILL.md YAML frontmatter parser & AST security scanner
│   ├── validators/            # Zod schemas for skills, manifests, and submissions
│   └── adapters/              # Multi-agent directory mappings and command generators
├── scripts/
│   ├── build-1000-catalog.js  # Catalog generator & metadata validator
│   └── expand-skills.js       # Capability expansion script
├── package.json
├── turbo.json
└── README.md

Package Descriptions

  • @domoskills/cli: The command-line client published as domoskills (invocable via npx domoskills). Handles file downloads, lockfile generation, security auditing, and environment doctor checks.
  • @domoskills/registry: In-memory search indexing, category trees, scoring algorithms, and curated skill metadata containing over 1,000 production skills.
  • @domoskills/skill-parser: Parses SKILL.md files with YAML frontmatter delimiters, extracts metadata, validates AST nodes, and runs security regex heuristics.
  • @domoskills/validators: Type-safe Zod schemas for skill manifests, frontmatter definitions, trust levels, and user submissions.
  • @domoskills/adapters: Filesystem routing and configuration markers for 7 different AI coding assistants and agent runtimes.
  • apps/web: Production web application built with Next.js 14, Tailwind CSS, Lucide icons, Zustand state management, and Framer Motion.

Supported Agent Target Ecosystems

DomoSkills provides an adapter layer supporting all modern AI coding agents:

Agent Target Flag Installation Path Configuration Marker Auto-Discovery Global User Path
Universal Standard --agent universal .agent/skills/ .agent/agent.yaml Yes ~/.agent/skills
Claude Code --agent claude .claude/skills/ .claude/config.json Yes ~/.claude/skills
OpenAI Codex / Agents --agent codex .agents/skills/ .agents/manifest.json Yes ~/.agents/skills
Cursor IDE --agent cursor .cursor/skills/ .cursorrules Yes ~/.cursor/skills
OpenCode Interpreter --agent opencode .opencode/skills/ .opencode/opencode.json Yes ~/.opencode/skills
GitHub Copilot --agent copilot .github/skills/ .github/copilot-instructions.md Yes ~/.github/skills
Gemini CLI / Antigravity --agent gemini .gemini/skills/ .gemini/config/skills Yes ~/.gemini/skills

How Agent Resolution Works

When you pass --agent <target>:

  1. DomoSkills determines the designated target directory (e.g. .cursor/skills/ or .claude/skills/).
  2. It verifies the destination directory and checks for configuration hooks.
  3. The skill package and its SKILL.md instruction file are extracted into <agent_dir>/<skill-name>/.
  4. The local domoskills.json lockfile is updated with the installed skill, checksum, and target agent.

CLI Command Reference

The domoskills CLI is fully executable with npx domoskills or after installing globally (npm install -g domoskills).

domoskills init

Initialize a domoskills.json configuration file in the current directory.

# Initialize with default universal target
npx domoskills init

# Initialize with a specific agent target
npx domoskills init --agent cursor
npx domoskills init --agent claude --force

Options:

  • -a, --agent <target>: Target agent ecosystem (universal, claude, codex, cursor, opencode, copilot, gemini). Default: universal.
  • -f, --force: Overwrite existing configuration.

domoskills search

Search skills in the open registry directly from your terminal.

# Search by keyword
npx domoskills search react

# Filter by category or agent compatibility
npx domoskills search graphql --category backend
npx domoskills search testing --agent cursor

Options:

  • -c, --category <category>: Filter results by category slug (frontend, backend, security, ai-ml, etc.).
  • -a, --agent <agent>: Filter by agent compatibility.

domoskills add

Add one or more agent skills to your project or global user profile.

# Install single skill
npx domoskills add react-performance

# Install multiple skills simultaneously
npx domoskills add nextjs-app-router tailwind-design-system --agent cursor

# Install globally in your user home directory
npx domoskills add owasp-agent-guardian --global

# Force re-installation
npx domoskills add fastapi-clean-architecture --force

Options:

  • -a, --agent <target>: Target agent ecosystem.
  • -g, --global: Install globally into ~/<agent>/skills.
  • -f, --force: Force re-installation even if already present.

domoskills install

Install all skills declared in your domoskills.json lockfile. Ideal for CI/CD pipelines and newly cloned repositories.

# Install everything declared in domoskills.json
npx domoskills install

# Force overwrite existing files
npx domoskills install --force

Options:

  • -a, --agent <target>: Override target agent ecosystem.
  • -f, --force: Force overwrite existing skills.

domoskills remove

Safely remove an installed skill from the project workspace and update the lockfile.

npx domoskills remove react-performance
npx domoskills rm nextjs-app-router --agent cursor

Options:

  • -a, --agent <target>: Target agent ecosystem.

domoskills list

Display all skills currently installed in your active workspace or agent folder.

npx domoskills list
npx domoskills ls --agent claude

domoskills doctor

Inspect your local environment, Node.js version, Git repository root, and detect active AI agent configurations (.cursor, .claude, .agent, .gemini, etc.).

npx domoskills doctor

Provides a diagnostic health checklist with actionable repair suggestions for any missing markers.


domoskills audit

Perform a local AST security audit on all installed skills in your project, scanning for quarantined scripts, unsafe patterns, or security advisories.

npx domoskills audit
npx domoskills audit --agent cursor

domoskills update

Check the remote registry and update installed skills to their latest verified versions.

# Update a specific skill
npx domoskills update react-performance

# Update all installed skills
npx domoskills update

The SKILL.md Specification

Every DomoSkills package contains a canonical SKILL.md file that encapsulates capability instructions, operational guidelines, constraints, and trigger words for AI agents.

YAML Frontmatter Schema

Each SKILL.md must begin with YAML frontmatter bounded by ---:

---
name: owasp-agent-guardian
description: Enforces OWASP Top 10 security standards, input sanitization, and dependency hardening for AI coding tasks.
version: 1.2.0
license: Apache-2.0
author: domoskills
compatibility:
  - universal
  - claude
  - cursor
  - gemini
  - codex
tags:
  - security
  - owasp
  - hardening
  - authentication
requiresEnv: []
dependencies: []
---

Supported Fields

Field Type Required Description
name string Yes Unique hyphenated skill identifier (e.g. react-performance).
description string Yes Concise summary of what capability this skill teaches the agent.
version string Yes Semantic version number (1.0.0).
license string Yes SPDX license identifier (MIT, Apache-2.0, BSD-3-Clause).
author string No Author or organization handle.
compatibility string[] No List of compatible agent targets (universal, claude, cursor, etc.).
tags string[] No Searchable keywords and capability categories.
dependencies string[] No Sub-skills or packages this skill depends upon.
requiresEnv string[] No Environment variables the skill may reference (audited during scan).

Recommended Markdown Structure

To ensure maximum retention and adherence by AI agents, structure your skill markdown as follows:

# Skill Name

## When to Trigger
Specify precise keywords, file patterns, or developer prompts that activate this skill:
- When writing SQL queries, authentication middleware, or API routes.
- Files matching `*.controller.ts`, `auth/*.ts`, or `api/**/*.py`.

## Core Guidelines & Best Practices
Provide concrete, prescriptive rules that the agent must adhere to:
1. Always validate incoming request payloads with Zod or Pydantic.
2. Use parameterized queries exclusively; forbid string concatenation in database calls.
3. Keep tokens out of version control and read from validated environment variables.

## Anti-Patterns & Pitfalls
Explicitly highlight what the AI coding agent must NEVER do:
- ❌ Do NOT use `eval()` or dangerous deserialization primitives.
- ❌ Do NOT bypass CSRF protection or disable CORS headers in production.

## Concrete Code Examples
Provide clear before/after code blocks demonstrating the desired implementation pattern.

Security Model & Auditing

AI coding assistants have elevated access to developers' file systems and terminals. DomoSkills treats security as a first-class citizen with multiple layers of defense:

The 4 Core Security Pillars

  1. Zero Automatic Execution: DomoSkills will NEVER execute arbitrary scripts or binaries downloaded from remote repositories during installation. Skills are purely declarative knowledge files and references.
  2. Path Traversal Guard: Every file in a package is validated with strict path normalization. Null bytes (\0), absolute path segments (/var, C:\), and parent directory traversals (../) are instantly rejected.
  3. Quarantine & Auditing: Compiled binaries (.exe, .dll, .bin, .so) and executable scripts (.sh, .bat, .py) are automatically quarantined in read-only mode and flagged with warning badges in both the UI and CLI audit logs.
  4. Secret & Pattern Scanning: AST-level static pattern matching scans skill markdown and companion files for hardcoded API credentials, private cryptographic keys, malicious pipes, and destructive shell commands.

100-Point Security Score & Trust Levels

Every skill is evaluated and assigned an objective security score:

  • 100 / 100: Baseline score for verified declarative Markdown instructions with valid metadata, validated licensing, and zero script presence.
  • -10 Points: Presence of executable script files (.sh, .bash, .py, .bat).
  • -25 to -30 Points: Suspicious shell patterns or credential structures detected in text.
  • -40 Points: Binary file artifacts (.bin, .so, .dylib, .exe).
  • -50 Points: Forbidden path characters or attempted traversal escapes.

Trust Tiers

Trust Tier Badge Color Criteria
Official Purple / Cyan Curated and maintained directly by the DomoSkills core team.
Verified Emerald Green Verified publisher, Security Score $\ge$ 80, no executable script files.
Community Blue Community contributed, Security Score $\ge$ 60, clean AST scan.
Experimental Amber / Red Security Score < 60 or unreviewed community submission.

Prohibited & Flagged AST Patterns

The parser flags the following dangerous patterns during registry ingestion and local auditing:

  • Shell Pipes: curl ... | bash or wget ... | sh
  • Destructive Deletion: rm -rf / or rm -rf ~ or rm -rf $HOME
  • Reverse Shells: Netcat reverse shell calls (nc -e /bin/sh)
  • Encoded Commands: Encoded PowerShell payloads (powershell -e ...)
  • Hardcoded Secrets: GitHub tokens (ghp_...), Slack tokens (xoxb-...), AWS keys (AKIA...)
  • Private Keys: -----BEGIN PRIVATE KEY----- or -----BEGIN RSA PRIVATE KEY-----
  • Obfuscated Code: eval(base64_decode(...))

Web Platform & Discovery Engine

The DomoSkills web application (apps/web) provides a rich, responsive interface for browsing, comparing, and configuring skills:

  • Explore Page (/explore):
    • Filter across 12 tech domains (Frontend, Backend, Security, DevOps, Cloud, AI/ML, etc.).
    • Search by title, tag, author, or description in milliseconds.
    • Multi-agent compatibility selector (Universal, Claude Code, Cursor, Codex, Copilot, Antigravity, OpenCode).
    • Density switch: Card Grid mode, Compact Table mode, or List mode.
    • Sort by Popularity, Rating, Security Score, or Alphabetical order.
  • Skill Detail Page (/skills/[slug]):
    • Full rendered markdown view of the skill's instructions.
    • Interactive AST Security Scorecard with itemized criteria and warning reports.
    • One-click copy installation commands for all 7 supported AI agents.
    • File tree browser showing companion reference files.
  • Skill Comparator Drawer:
    • Compare up to 4 skills side-by-side.
    • View comparative metrics: security scores, repository stars, file counts, license types, and trigger keywords.
  • Skill Stack Cart:
    • Collect skills as you browse with animated fly-to-cart micro-interactions.
    • Generate a single compound terminal command to install your entire stack at once:
      npx domoskills add react-performance tailwind-tokens owasp-agent-guardian --agent cursor
    • Export your stack directly to domoskills.json.
  • Environment Doctor (/doctor):
    • Web diagnostic interface verifying local agent configurations and markers.
  • Skill Submission Studio (/submit):
    • In-browser markdown and frontmatter editor with live validation against Zod schemas.
    • Pre-flight security scoring before submitting to the open registry.
  • Command Palette (Cmd+K / Ctrl+K):
    • Instant modal for fuzzy searching the catalog, jumping between pages, and executing platform actions.

Curated Starter Packs

DomoSkills includes pre-configured capability stacks for common engineering initiatives:

🛡️ The OWASP Security Guardian Stack

npx domoskills add owasp-agent-guardian secure-auth-patterns secrets-hardening --agent universal

Arm your AI assistant with OWASP Top 10 rules, JWT/session security best practices, and zero-leakage secret handling.

⚛️ The Next.js 14 & React Performance Stack

npx domoskills add nextjs-app-router react-performance tailwind-design-system --agent cursor

Equip Cursor or Claude with deep knowledge of React Server Components, hydration optimization, and design tokens.

🐍 Python Clean Architecture & FastAPI Stack

npx domoskills add fastapi-clean-architecture pydantic-v2-patterns pytest-asyncio-suite --agent claude

Enforce strict domain-driven design, Pydantic v2 data validation, and asynchronous testing patterns.

🤖 AI Agent & RAG Pipeline Stack

npx domoskills add rag-pipeline-architect vector-db-optimizer llm-prompt-evaluator --agent opencode

Optimize hybrid search retrieval, chunking strategies, embeddings, and automated prompt evaluation harnesses.


domoskills.json Lockfile

When installing skills, DomoSkills automatically creates or updates a reproducible domoskills.json lockfile in your project root:

{
  "version": 1,
  "agent": "universal",
  "skills": [
    {
      "name": "react-performance",
      "source": "domoskills/official-agent-skills",
      "version": "1.4.2",
      "commit": "7f9a12c",
      "installedAt": "2026-08-30T12:00:00.000Z"
    },
    {
      "name": "owasp-agent-guardian",
      "source": "domoskills/official-agent-skills",
      "version": "1.2.0",
      "commit": "b3e412f",
      "installedAt": "2026-08-30T12:00:05.000Z"
    }
  ]
}

Reproduce your entire capability stack on any developer machine or CI container:

npx domoskills install

Local Development

DomoSkills uses pnpm and Turborepo to manage packages and dependencies.

Prerequisites

  • Node.js >= 18.0.0
  • pnpm >= 9.0.0

Step-by-Step Setup

# Clone the repository
git clone https://github.com/darknecrocities/DomoSkills.git
cd DomoSkills

# Install all workspace dependencies
pnpm install

# Build all packages & web application
pnpm build

# Run unit and integration tests across all packages
pnpm test

# Launch web dev server (http://localhost:3000)
pnpm dev

Useful Monorepo Commands

Command Description
pnpm build Compiles TypeScript for all 5 packages and creates the Next.js production build.
pnpm test Runs Vitest across @domoskills/adapters, @domoskills/registry, @domoskills/skill-parser, and @domoskills/validators.
pnpm dev Starts the Next.js development server with hot reloading.
pnpm lint Runs Next.js ESLint checks across the codebase.

Contributing

Contributions from the developer and AI agent communities are warmly welcomed!

  1. Fork the Repository: Create a personal fork on GitHub.
  2. Create a Feature Branch: git checkout -b feat/my-new-skill-or-feature.
  3. Write Tests: Add Vitest test cases for any new parsers, adapters, or validators.
  4. Verify Standards: Ensure pnpm build and pnpm test pass with zero errors.
  5. Submit a Pull Request: Provide a concise summary of the changes and any new skills introduced.

Submitting a New Skill

You can submit a skill via pull request in packages/registry/src/data/skills.json or by using the visual submission portal at http://localhost:3000/submit.

Ensure your SKILL.md:

  • Contains valid YAML frontmatter conforming to @domoskills/validators.
  • Has an open-source license (MIT, Apache-2.0, BSD-3-Clause).
  • Scores $\ge 80$ on the AST security scanner.
  • Contains clear trigger conditions and actionable best practices.

License

Licensed under the MIT License.

Curated skills preserve their original open-source repository licenses (MIT, Apache-2.0, BSD).

About

DomoSkills is a developer-native discovery engine and CLI installer for AI Agent Skills. Discover modular capabilities, curate your project's skill stack, and install directly into your AI coding assistant workspace in a single command.

Topics

Resources

Stars

12 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages