cyber.hendryadrian.com
Security News · opensourcemalware.com

Podcast Covers GitHub Account Protections, Malicious npm Packages and FBI TeamPCP Notice

· Original article ↗

Summary

The episode discusses GitHub account-security changes, six allegedly malicious npm packages targeting AI companies, an FBI FLASH notice on TeamPCP, and security concerns as AI IDEs increasingly use Open VSX.

Key points

  • npm introduced a 72-hour read-only hold for high-impact accounts after an email change or use of a 2FA recovery code, intended to slow account takeovers.
  • The hosts warn the hold could also delay legitimate maintainers from recovering compromised accounts and removing malicious package changes.
  • GitHub Enterprise added self-service revocation of user credentials, including tokens, SSH keys and SSO authorizations; the hosts question whether it covers personal tokens outside the enterprise boundary.
  • The hosts report that a security researcher published six npm packages targeting users of OpenAI, Anthropic, Vercel and Ollama. They say the packages collected data from .git and .ssh but did not directly exfiltrate credentials.
  • The FBI issued a FLASH notice about TeamPCP’s behavior and indicators of compromise; the episode points listeners to the notice for details.
  • The hosts say AI IDEs including Cursor and Windsurf use Open VSX, making the extension registry a more attractive target and increasing the importance of trusted distribution and security controls.

Article Details

Event Type
Cybersecurity news roundup covering GitHub account-security updates, malicious npm packages, and an FBI FLASH notice on TeamPCP.
Impact
The reported npm packages collect data from .git and .ssh directories while targeting users of AI-related companies; the speakers say they do not directly exfiltrate credentials. GitHub’s account protections may delay legitimate maintainer recovery, and Enterprise credential revocation may not cover credentials created outside the enterprise boundary. The FBI FLASH provides information on TeamPCP behavior and indicators, but none are listed in the article.

Threat Actors

TeamPCPNamed in the article as the subject of an FBI FLASH notice covering its behavior and indicators; a speaker says TeamPCP may be one person, but provides no identity.

Vendors

Anthropicfound six npm packages published by a researcher at a stealth cybersecurity startup, targeting OpenAI, Anthropic, Vercel, and Ollama users. The packages avoid exfiltrating credentials directly but do pull data fromGitHubGitHub’s two new account protection features: NPM added a 72-hour read-only lockout for high-impact accounts triggered by an email change or 2FA recovery code use, aimed at slowing account takeovers. Separately, GitHubOpenAIPaul found six npm packages published by a researcher at a stealth cybersecurity startup, targeting OpenAI, Anthropic, Vercel, and Ollama users. The packages avoid exfiltrating credentials directly but do pullSnykcrosses an ethical line for security research. He draws a direct parallel to the 2025 incident where a Snyk researcher published malicious packages targeting Cursor.Vercelnpm packages published by a researcher at a stealth cybersecurity startup, targeting OpenAI, Anthropic, Vercel, and Ollama users. The packages avoid exfiltrating credentials directly but do pull data from .git and

Products

Cursora direct parallel to the 2025 incident where a Snyk researcher published malicious packages targeting Cursor.GitHub Enterpriseby an email change or 2FA recovery code use, aimed at slowing account takeovers. Separately, GitHub Enterprise rolled out self-service credential revocation, letting enterprise owners revoke tokens, SSH keys,npmGitHub’s two new account protection features: NPM added a 72-hour read-only lockout for high-impact accounts triggered by an email change or 2FA recovery code use, aimed at slowing account takeovers. Separately, GitHubOllamapublished by a researcher at a stealth cybersecurity startup, targeting OpenAI, Anthropic, Vercel, and Ollama users. The packages avoid exfiltrating credentials directly but do pull data from .git and .ssh, whichOpen VSXand Paul will be joined by Mikael Barbero, Head of Security at the Eclipse Foundation, to talk about Open VSX. The draw: AI IDE platforms like Cursor and Windsurf use Open VSX instead of the official VS CodeWindsurfSecurity at the Eclipse Foundation, to talk about Open VSX. The draw: AI IDE platforms like Cursor and Windsurf use Open VSX instead of the official VS Code marketplace, which is turning it into a more attractive

Related Articles