Skip to content

Back to the catalog

♫Why♥Not♪ (@Python_s_) on X
x.comadvisory

♫Why♥Not♪ (@Python_s_) on X

Alert summarizing Socket Threat Research's disclosure that a malicious version of the Tensorlake npm SDK (an AI-agent infrastructure package) was published with a preinstall hook delivering Shai-Hulud/ChainDrop credential-stealing malware. The payload targets GitHub, npm, AWS, Kubernetes and Vault credentials, persists via a gh-token-monitor mechanism, and spreads through compromised publishing accounts. It lists IOCs (lib/setup.mjs, lib/Math_Symbol.js) and warns to remove the persistence before revoking stolen GitHub tokens, since revocation triggers a destructive routine. Socket detected the release about 11 minutes after publication.

TL;DR Tensorlake npm SDK trojanized with Shai-Hulud/ChainDrop preinstall credential stealer; remove gh-token-monitor persistence before revoking GitHub tokens or a destructive routine fires.
Document

What it is

A secondary alert on X (not Socket's own post) dated October 8, 2026, reporting that tensorlake@0.5.144 on npm is malicious. Tensorlake is an SDK used in AI-agent infrastructure. The alert attributes the disclosure to Socket Threat Research and classes the malware as ChainDrop / Shai-Hulud.

How it works

  • The malicious version runs code at install time through a preinstall hook.
  • The payload steals GitHub, npm, AWS, Kubernetes and Vault credentials, sets up persistence, and propagates through compromised publishing accounts.
  • The persistence indicator is gh-token-monitor.
  • The destructive routine fires when stolen GitHub tokens are revoked.

Preconditions / Affected

  • Only version 0.5.144 is named. The alert gives no safe version or fixed release.
  • Any development machine or CI/CD runner that installed that version is in scope, because the hook runs at install time.
  • The number of affected installations is stated as unknown.

What to do

Check for the version and the indicators first:

  • Package: tensorlake@0.5.144
  • Files: lib/setup.mjs, lib/Math_Symbol.js
  • Persistence: gh-token-monitor

Then follow the alert's order of response:

  1. Block the compromised version and investigate dev and CI/CD environments that may have installed it.
  2. Isolate affected hosts.
  3. Remove the token-monitor persistence before revoking any stolen GitHub tokens, because revocation triggers the destructive routine.
  4. Rotate exposed secrets only after containment.
  5. Rebuild compromised environments from trusted sources.

Caveats

  • The alert gives no removal commands for the persistence. It doesn't say where gh-token-monitor lives or what the destructive routine deletes. The Socket blog link is truncated, so get those details from Socket's original write-up before acting.
  • The "very high" confidence rating is the alert author's own, resting on Socket's analysis. I haven't checked it against the primary source.
  • Socket's detection about 11 minutes after publication doesn't bound exposure. Anyone who installed during that window, or before the package was pulled, is still affected.

Organisations named here

  • gh-token-monitor
  • lib/Math_Symbol.js
  • lib/setup.mjs
  • Shai-Hulud

Technologies named here

  • Tensorlake npm SDK
  • Socket Threat Research
  • Socket blog post
  • Credentials at risk
  • npm
  • tensorlake
  • Socket
  • GitHub
  • AWS
  • Kubernetes

Techniques named here

  • Software supply-chain compromise of an npm package
  • Credential theft
  • Persistence
  • Propagation through compromised publishing accounts
  • Destructive routine triggered by token revocation
  • Malicious files

Threat actors named here

  • Shai-Hulud
  • ChainDrop