VYPR

gem · Malicious package advisory

Malware

crypti-toolbox

MAL-2026-17593

Malicious code in crypti-toolbox (RubyGems)

Details

RubyGems runs this gem's native extension script `ext/crypti-toolbox/extconf.rb` during `gem install`. The script exits without action on CI runners and sandboxes, based on CI environment variables, the hostname, the username and the working directory. It continues if the system uptime is over 30 minutes and the home directory holds `.ssh`, `.gitconfig`, `.npmrc`, `.gem/credentials` or `.bundle`. It then decodes a Base64 Ruby payload and forks twice into a detached process. After 20 to 40 minutes the payload downloads `http://45.138.12.177:8092/wgkit.tar.gz` to `/tmp/.w1.tgz` with `curl`, extracts it to `/tmp/.w1/`, runs `wg_install.sh` with `bash` and deletes both. The payload stores the URL XOR-encoded with the key `usv\x9a`. The Ruby code in `lib/` is a decoy. The RubyGems account `reqthrottle_3474` published this gem and 41 others with the same loader on 2026-10-05.

---
_-= Per source details. Do not edit below this line.=-_

## Source: amazon-inspector (5ec4d5d5915335d90b2f18b45b050936fa5d2db9d176d7d081bd8c292c7c6795)
On `gem install`, the native extension script ext/crypti-toolbox/extconf.rb runs an environment-fingerprinting routine that bails out on CI runners, ephemeral/sandbox hostnames, generated or analysis-shaped usernames, analysis path prefixes (/opt/rubygems, /tmp, /workspace), low machine uptime, and hosts lacking developer artifacts (~/.ssh, ~/.gitconfig, ~/.gem/credentials). On hosts that pass those checks, the script base64-decodes a Ruby snippet and passes it to eval inside a double-forked, setsid-detached child that sleeps 20-40 minutes and then curls a tarball from http://45.138.122.177:8092/wgkit.tar.gz to /tmp/.w1.tgz, extracts it to /tmp/.w1, and executes the embedded wg_install.sh under bash. The destination URL is hidden as a hex string XORed with a 4-byte key, and the shell command is wrapped in a base64-encoded eval payload. There is no pin, hash, or signature on the fetched content, the destination is a bare IP unrelated to any declared publisher, and the sandbox-evasion and daemonization logic exist only to attack real developer workstations while hiding from analysis environments.

## Source: ossf-package-analysis (5698e127e62777cef2293c537c1754d933a4be1cdd351111bb9cc0f3e3710e96)
The OpenSSF Package Analysis project identified 'crypti-toolbox' @ 1.0.0 (rubygems) as malicious.

It is considered malicious because:

- The package executes one or more commands associated with malicious behavior.

Compromised versions (1)

  • 1.0.0

Any computer that installed or ran a compromised version should be considered fully compromised. Rotate every secret on that machine from a clean environment.