CrawSecureCrawSecure
Privacy-first Β· Offline Β· Open source
OpenClaw Skills

Scan skills. Keep your code.

Detect dangerous patterns in ClawHub skills before you install them. Analysis runs entirely on your machine.

β€”skills scanned
13detection rules
.js .ts .shfile types
bash
$ npx crawsecure ./my-skill

β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”
β”‚  CrawSecure v2  Β·  @username  [FREE]             β”‚
β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜

Target: ./my-skill

🚨 Security signals found: 3

  πŸ”΄ [HIGH]   Detected eval() usage β€” src/index.js
  🟑 [MEDIUM] child_process detected β€” src/utils.js
  🟒 [LOW]    process.env access β€” src/config.js

πŸ“Š Risk score: 8500 β†’ HIGH

  Scan saved  β†’  crawsecure.com/dashboard/abc123
  Free Β· 3 / 10 scans this month

How it works

Three steps, zero uploads, instant results.

01

Drop files or run the CLI

Drag project files into the browser scanner, or run `npx crawsecure .` from any directory.

02

Analysis runs locally

The engine reads file contents in memory and applies 13 security rules. Nothing leaves your device.

03

Get your security report

See which rules fired, severity levels, and an overall risk score β€” instantly.

What we detect

13 stable rule IDs β€” all public, all auditable.

See source β†—
rm-rfDestructive rm -rf command
evalDynamic code execution via eval()
execProcess execution via exec()
ssh-dirReferences .ssh directory
id-rsaReferences SSH private key
child-processExecutes system commands
spawnChild process via spawn()
curlNetwork request via curl
wgetFile download via wget
dotenvReferences .env file (secrets)
walletReferences wallet file
credentialsReferences credentials file
process-envAccess to environment variables
High riskMedium riskLow risk

Your code never touches our servers. Ever.

Open DevTools β†’ Network tab while running a scan. You will see zero outbound requests during analysis. If you choose to save a scan, you'll see exactly what is sent β€” a handful of numbers, nothing more.

Read our privacy commitment
  • Code processed entirely in memory β€” never written to disk
  • Only aggregated numbers saved (if you choose to)
  • No file names, no paths, no code snippets β€” ever
  • Fully auditable: open source, MIT license

Frequently asked questions

Ready to scan?

No account required. Sign in only when you want to save your history.