research

Fail

Audited by Socket on Oct 17, 2026

3 alerts found:

Malwarex3
MalwareHIGH
references/last30days/scripts/lib/cookie_extract.py

This module is a local browser cookie extractor. It searches for Firefox (including WSL access to Windows profiles), copies cookies.sqlite to a temp file, queries moz_cookies for a specified domain and cookie names, and returns the cookie values to the caller—behavior strongly aligned with session/cookie theft. It further orchestrates extraction for other browsers on macOS via helper modules. No clear protective measures or legitimate-use safeguards are present in this fragment.

Confidence: 41%Severity: 85%
MalwareHIGH
references/last30days/scripts/lib/chrome_cdp.py

This module is a CDP-based browser cookie harvester that extracts and returns a targeted authentication cookie pair (auth_token and ct0) associated with x.com/twitter.com from a live Chrome/Chromium session. The mechanism (Network.getAllCookies) and the specific cookie targeting strongly align with credential/session theft rather than benign functionality. Risk is elevated further by the ability to override the CDP endpoint via BROWSER_CDP_URL and by the custom low-level WebSocket implementation. The only meaningful uncertainty is whether this code is used in a consented/authorized context, and the fragment appears truncated/typoed at the end, but the sensitive behavior is fully observable in the provided logic.

Confidence: 78%Severity: 95%
MalwareHIGH
references/last30days/scripts/lib/chrome_cookies.py

This module contains a highly suspicious, targeted capability to harvest and decrypt Chromium-family browser cookies on macOS. It enumerates and copies browser cookie SQLite databases, obtains the necessary decryption passphrase from the macOS Keychain via `security`, decrypts v10 encrypted cookie values using OpenSSL, and returns plaintext cookie/session material for caller-selected domains and cookie names. Even without visible exfiltration in this snippet, the implemented behavior aligns strongly with cookie/session theft tooling and presents a critical privacy/security risk if included in a software supply chain.

Confidence: 90%Severity: 100%
Audit Metadata
Analyzed At
Oct 17, 2026, 01:30 AM
Package URL
pkg:socket/skills-sh/lev-os%2Fagents%2Fresearch%2F@51c088c91db2417e711cf575c3a6323b7f05ec3f9fb0e8f6c4d1f780179892f2