Skip to content

Installation Guide

Comprehensive installation instructions for Skill Engine on all supported platforms.

System Requirements

Minimum Requirements

  • OS: Linux, macOS 11+, or Windows 10+ (with WSL2)
  • RAM: 512MB available memory
  • Disk: 50MB for binary + storage for skills

No Rust compiler needed! Skill Engine is distributed as a pre-compiled binary.

Optional Requirements

  • Docker: For running Docker-based skills (containerized tools)
  • Rust: Only needed if building from source (development)

Installation Methods

The easiest way to install Skill Engine:

bash
curl -fsSL https://dqkbk9o7ynwhxfjx.public.blob.vercel-storage.com/install.sh | sh

This script:

  • Detects your OS and architecture automatically
  • Downloads the pre-compiled binary (no Rust needed!)
  • Installs to ~/.skill-engine/bin/skill
  • Adds to your PATH (with permission)

Verify installation:

bash
skill --version
# Output: skill 0.3.0

Install specific version:

bash
curl -fsSL https://dqkbk9o7ynwhxfjx.public.blob.vercel-storage.com/install.sh | SKILL_VERSION=0.2.0 sh

Non-interactive install (for CI/CD):

bash
# Won't modify PATH, won't ask for confirmation
SKILL_NO_MODIFY_PATH=1 curl -fsSL https://dqkbk9o7ynwhxfjx.public.blob.vercel-storage.com/install.sh | sh

Custom install directory:

bash
SKILL_INSTALL_DIR=/opt/skill curl -fsSL https://dqkbk9o7ynwhxfjx.public.blob.vercel-storage.com/install.sh | sh

Method 2: Build from Source (Optional)

For developers who want to build from source or contribute:

Requirements: Rust 1.75.0 or later (rustup.rs)

bash
# Install Rust (if not already installed)
curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh

# Clone repository
git clone https://github.com/kubiyabot/skill.git
cd skill

# Build and install
cargo install --path crates/skill-cli

Platform-Specific Notes

Linux

The one-liner install works on all Linux distributions. No additional setup needed!

Optional: Install Docker (for Docker-based skills):

Ubuntu/Debian

bash
sudo apt-get update
sudo apt-get install docker.io
sudo systemctl start docker
sudo usermod -aG docker $USER

Fedora/RHEL

bash
sudo dnf install docker
sudo systemctl start docker
sudo usermod -aG docker $USER

Arch Linux

bash
sudo pacman -S docker
sudo systemctl start docker
sudo usermod -aG docker $USER

macOS

The one-liner install works on all macOS versions 11+.

Optional: Install Docker (for Docker-based skills):

bash
# Download Docker Desktop
# Visit: https://www.docker.com/products/docker-desktop

# Or via Homebrew
brew install --cask docker

Windows (WSL2)

The one-liner install works in WSL2 Ubuntu/Debian.

Optional: Install Docker:

  1. Download Docker Desktop for Windows with WSL2 backend
  2. Visit: https://www.docker.com/products/docker-desktop

Claude Code Setup

If you're using Claude Code, set up MCP integration automatically:

bash
skill claude setup

This command:

  • Finds your skill binary automatically
  • Creates/updates .mcp.json in your current directory
  • Configures the MCP server for Claude Code

For global setup (use across all projects):

bash
skill claude setup --global

Verify integration:

bash
skill claude status

Output:

✓ Claude Code integration configured
  Location: /path/to/project/.mcp.json
  Server name: skill-engine
  Binary: /Users/you/.skill-engine/bin/skill

Restart Claude Code and you're ready to use skills through Claude!

See the Claude Code Integration Guide for advanced configuration and tips.

Shell Completion

Enable command-line completion for your shell:

Bash

bash
# Generate completion script
skill completions bash > ~/.local/share/bash-completion/completions/skill

# Or add to ~/.bashrc
echo 'eval "$(skill completions bash)"' >> ~/.bashrc
source ~/.bashrc

Zsh

bash
# Generate completion script
skill completions zsh > ~/.zfunc/_skill

# Add to ~/.zshrc
echo 'fpath=(~/.zfunc $fpath)' >> ~/.zshrc
echo 'autoload -Uz compinit && compinit' >> ~/.zshrc
source ~/.zshrc

Fish

bash
# Generate completion script
skill completions fish > ~/.config/fish/completions/skill.fish

Configuration

Default Configuration

Create ~/.config/skill-engine/config.toml:

toml
# Default manifest file location
manifest_path = ".skill-engine.toml"

# Enable execution history
history_enabled = true
history_limit = 1000

# HTTP server settings
[server]
host = "127.0.0.1"
port = 3000
ui_enabled = true

# Docker runtime settings
[docker]
default_memory = "2g"
default_cpus = "4"
network = "bridge"

# WASM runtime settings
[wasm]
max_memory_pages = 256  # 16MB
max_execution_time_ms = 30000

Environment Variables

Skill Engine respects these environment variables:

bash
# Configuration file location
export SKILL_ENGINE_CONFIG=~/.config/skill-engine/config.toml

# Manifest file name
export SKILL_ENGINE_MANIFEST=.skill-engine.toml

# Enable debug logging
export RUST_LOG=skill=debug

# Docker host
export DOCKER_HOST=unix:///var/run/docker.sock

# HTTP server port
export SKILL_ENGINE_PORT=3000

Add to ~/.bashrc or ~/.zshrc to make permanent.

MCP Server Configuration

For Claude Code and other MCP-compatible agents:

Claude Code

Create or edit ~/.config/claude/mcp.json:

json
{
  "mcpServers": {
    "skill-engine": {
      "command": "skill",
      "args": ["mcp"],
      "env": {
        "RUST_LOG": "info"
      }
    }
  }
}

Cursor

Create or edit ~/.cursor/mcp.json:

json
{
  "mcpServers": {
    "skill-engine": {
      "command": "skill",
      "args": ["mcp"]
    }
  }
}

Verification

Verify your installation is working correctly:

bash
# Check version
skill --version

# List available skills
skill list

# Check system info
skill info

# Start HTTP server
skill serve &

# Test API endpoint
curl http://localhost:3000/api/health

# Stop server
pkill skill

Troubleshooting

Command Not Found

Issue: skill: command not found

Solution: Ensure ~/.cargo/bin is in your PATH:

bash
# Add to ~/.bashrc or ~/.zshrc
export PATH="$HOME/.cargo/bin:$PATH"
source ~/.bashrc  # or source ~/.zshrc

Docker Permission Denied

Issue: permission denied while trying to connect to Docker

Solution: Add user to docker group:

bash
sudo usermod -aG docker $USER
newgrp docker

# Or restart system

Rust Version Too Old

Issue: requires rustc 1.75.0 or newer

Solution: Update Rust:

bash
rustup update stable
rustc --version

Build Errors

Issue: Compilation errors during cargo install

Solution: Update Rust and dependencies:

bash
rustup update stable
cargo clean
cargo build --release

macOS Gatekeeper Warning

Issue: "skill cannot be opened because it is from an unidentified developer"

Solution: Allow the application:

bash
# Remove quarantine attribute
xattr -d com.apple.quarantine /usr/local/bin/skill

# Or right-click → Open in Finder

Uninstallation

Remove Binary

bash
# If installed via cargo
cargo uninstall skill-cli

# If installed manually
sudo rm /usr/local/bin/skill

Remove Configuration

bash
# Remove config directory
rm -rf ~/.config/skill-engine

# Remove completion scripts
rm ~/.local/share/bash-completion/completions/skill  # Bash
rm ~/.zfunc/_skill  # Zsh
rm ~/.config/fish/completions/skill.fish  # Fish

Remove Docker Images

bash
# List Skill Engine related images
docker images | grep skill

# Remove specific image
docker rmi <image-name>

# Or remove all unused images
docker image prune -a

Updating

Update via Cargo

bash
cargo install skill-cli --force

Update from Source

bash
cd skill
git pull
cargo install --path crates/skill-cli --force

Next Steps

Support

Released under the Apache-2.0 License.