Getting Started
Quick Start
Install Sigil, scan your first package, and understand the verdict — all in under 60 seconds.
1. Install
Pick your preferred method. All three install the same CLI binary.
# curl (recommended)
curl -fsSL https://sigilsec.ai/install.sh | sh
# Homebrew
brew install nomarj/tap/sigil
# npm
npm install -g @nomarj/sigilThe curl and Homebrew installers cover macOS and Linux. On Windows, download sigil-windows-x64.zip from GitHub Releases and add it to your PATH.
2. Verify installation
sigil --version
# sigil 1.3.63. Scan a package
Sigil quarantines the target into a temporary directory, runs all eight scan phases, and outputs a verdict. Nothing touches your working environment until you approve.
# Scan an npm package
sigil npm lodash
# Scan a pip package
sigil pip requests
# Scan a git repo
sigil clone https://github.com/example/repo
# Scan a local directory
sigil scan ./my-project4. Read the verdict
Sigil outputs a risk score and verdict for every scan. The verdict tells you exactly what patterns were detected and the associated risk level.
┌─────────────────────────────────────────┐
│ SIGIL SCAN COMPLETE │
│ │
│ Target: npm:lodash@4.17.21 │
│ Score: 0 │
│ Verdict: ● LOW RISK │
│ │
│ Phases: │
│ ✓ Install Hooks — none found │
│ ✓ Code Patterns — none found │
│ ✓ Network / Exfil — none found │
│ ✓ Credentials — none found │
│ ✓ Obfuscation — none found │
│ ✓ Provenance — 1,200+ commits │
│ ✓ Prompt Injection — none found │
│ ✓ Skill Security — none found │
│ │
│ Action: pending — run sigil approve │
└─────────────────────────────────────────┘5. Set up shell aliases
Make security the default. These aliases intercept your normal commands and route everything through Sigil's quarantine-first workflow.
# Appends the aliases to your .bashrc / .zshrc
sigil setup shell
# Now these commands auto-quarantine and scan:
gclone https://github.com/example/repo
safepip requests
safenpm expressgit clone, pip install, and npm install is automatically quarantined and scanned. Nothing changes about your workflow except that nothing runs until you approve it.Next steps
- CLI Reference — all commands, flags, and output formats
- Configuration — custom policies, thresholds, and ignore rules
- MCP Integration — let AI agents scan before they install
Need help?
Ask a question in GitHub Discussions or check the troubleshooting guide.