Open-source · Free to use

Scan before you clone

SafeClone scans any GitHub repo for hardcoded secrets, vulnerable dependencies, and dangerous install scripts — before a single line of code touches your machine.

4
Scanners in parallel
10
Package ecosystems
<30s
Average scan time
terminal
$safeclone https://github.com/someorg/repo
🛡 SafeClone — scanning in sandbox...
✓ No secrets found
✓ No vulnerable dependencies
✓ No dangerous scripts
Verdict: SAFE ✓
Cloning into repo...
done.

Powered by trusted security tools

TruffleHog
OSV.dev
Docker Sandbox
Static Analysis

Three steps to safe cloning

SafeClone does all the heavy lifting on a remote server. Your machine stays clean the entire time.

Step 01

Paste a repo URL

Run safeclone <url> from your terminal. That's the entire workflow — no config, no project setup, no dependencies.

Step 02

We scan in a sandbox

Our server spins up an isolated Docker container, clones the repo inside it, and runs all three scanners in parallel. Nothing executes on your machine.

Step 03

Clone with confidence

Get a clear verdict in seconds. Safe repos clone automatically. Dangerous ones show you exactly what was found and prompt you before proceeding.

Four scanners, one command

Every scan runs all four checks simultaneously so you get a complete picture in one go.

Hardcoded secrets

Runs TruffleHog against the cloned filesystem. Only verified, active credentials are reported — no noise from already-rotated keys or test fixtures.

API keys Tokens Credentials Private keys

Vulnerable dependencies

Queries the OSV.dev database with exact version pinning across 10 package ecosystems. Only CVEs that affect the specific version declared in the repo are flagged.

npm PyPI Go Cargo Maven RubyGems Packagist NuGet Pub Hex

Dangerous install scripts

Statically analyzes install-time hooks in package manifests for patterns used in supply chain attacks — before any code runs anywhere.

eval / exec curl | sh base64 rm -rf process.env

AI prompt injection

Scans all markdown files for hidden instructions designed to hijack AI coding assistants — Copilot, Cursor, Claude — when they read the repo. Priority targets include CLAUDE.md, .cursorrules, and AGENTS.md.

CLAUDE.md .cursorrules AGENTS.md ignore previous instructions system prompt override

How the verdict is determined

SAFE — clones automatically

No secrets, no vulnerable deps, no dangerous scripts detected. SafeClone proceeds with git clone without prompting.

WARNING — prompts before cloning

Vulnerable dependencies or suspicious scripts found, but no active secrets. You see the full report and choose whether to proceed.

DANGEROUS — aborts by default

Active, verified secrets detected in the repository. SafeClone stops and shows you exactly what was found. Use --force to override.

Install in seconds

One binary. No runtime dependencies. No configuration needed.

Install via Homebrew
$brew install altship-hq/safeclone/safeclone
Use it
$safeclone https://github.com/some/repo
Add Scoop bucket
$scoop bucket add altship-hq https://github.com/altship-hq/scoop-safeclone
Install
$scoop install safeclone
Use it
$safeclone https://github.com/some/repo

Useful flags

--forceclone without prompting even if issues are found
--dirspecify a custom destination folder name

Never clone blind again.

Free, open-source, and takes 30 seconds to install.

Get started free