Sid Techno
HomePricingPortfolioContactSign in+1 (725) 465-8325WhatsAppGet Your Website
Sid Techno

Web & mobile app development agency. Websites, e-commerce stores, and custom software — built by the team behind DeployBase, TradeLeap, ClearAgent, and DeskLeap.

30 N Gould St, Ste NSheridan, WY 82801United States+1 (725) 465-8325

Stay Updated

Get product updates, new features, and hosting tips.

Products

  • DeployBase
  • TradeLeap
  • ClearAgent
  • DeskLeap

Services

  • CMS Development
  • Shopify Websites
  • React Native Apps
  • Laravel Development
  • Graphic Design
  • Wix Websites

Company

  • About Us
  • Pricing
  • Blog
  • Portfolio
  • Testimonials
  • Custom Solutions
  • Careers
  • Request a Quote
  • Contact
  • Client Portal
  • Sign In

Legal

  • Terms of Service
  • Privacy Policy
  • Refund Policy

© 2026 Sid Techno LLC. All rights reserved.

TermsPrivacyRefund
Home/Blog/Correct Publisher Name, Malicious Build: The Cursor Marketplace Extension Incident
October 31, 2026

Correct Publisher Name, Malicious Build: The Cursor Marketplace Extension Incident

On 29 October 2026 Cursor's extension marketplace served a malicious build under a real, trusted publisher name for about eleven hours. What the build did, why it partly failed, and why a publisher name is an identity claim, not a guarantee of what you install.

By Sid Techno

Update, 2 November 2026: Microsoft has published its own analysis of the malicious build. It names the files the persistence leaves behind and confirms which distribution channels were not affected. See what Microsoft added below.

On 29 October 2026 Cursor published a security notice about its extension marketplace. For roughly eleven hours that day, from about 2:27am to 1:35pm Pacific time, the Cursor VS Code Extension Marketplace served "a maliciously modified version of Anthropic's Claude Code extension", versions 2.1.312 and 2.1.313 for Windows and macOS. Cursor is explicit that "These malicious versions of the extension were not developed by Anthropic."

This is not a story about a bug in an editor or a poisoned dependency. It's about the distribution channel itself: a marketplace served a malicious build under a real, trusted publisher name. That makes it a useful lesson for anyone who installs extensions, which is nearly every developer.

What the malicious build did

Cursor and Phorion, a security company that sells endpoint detection software and published a technical analysis on 30 October, describe the same behaviour:

  • It ran on load. According to Phorion, the malicious code was prepended to the extension's main file and executes as soon as the editor loads it: "The user doesn't need to interact with the extension for the payload to fire."
  • It planted an SSH key. It added an attacker-controlled key to the user's ~/.ssh/authorized_keys, under the comment svc-telemetry.
  • It tried to open a way back in. It attempted to enable the local SSH service and hold open a reverse SSH tunnel to the attacker's server, and set up persistence so the tunnel would come back after a restart.
  • It reported on the machine. Hostname, user, OS, Git email, public IP, and the names (not the values) of environment variables that look like keys or tokens.

Why it was less bad than it could have been

The attack was partly broken by its own mistakes. Cursor says the step that turns on the local SSH service would fail "Outside of highly unusual circumstances where Cursor was started with administrator privileges, due to errors in the malicious code". Phorion found one of the reasons: on a Mac, the script checks whether Remote Login is off without the permissions needed to read that setting, gets an error message back, and so never tries to turn it on. In Phorion's words, "On a Mac that doesn't already have SSH running, that mistake breaks the inbound access path".

Cursor also worked with the hosting provider to take the attacker's server offline. It removed the versions, shut the whole marketplace down as a precaution, rotated its keys and rebuilt it. It says it "confirmed no other extensions were tampered with", and that it has "no evidence that other versions of the extension or the official distribution are affected."

But "partly broken" is not "harmless". Cursor is direct about what still happened on affected machines: "The extension added the attacker's SSH key, installed persistence, and sent host details to the attacker." And it warns that "Uninstalling the extension may not be sufficient", because the persistence scripts keep running on their own.

The lesson: a publisher name is not a guarantee

Every extension marketplace shows you who published an extension. That's an identity claim: it tells you which account the listing belongs to. It doesn't, by itself, prove that the file you download is the file that publisher built. Here the name on the listing was the real one, and the build was not.

Two things make that matter more for extensions than for most software:

  • Extensions run with your editor's permissions. That usually means your files, your source code, your SSH keys and your shell, often the most sensitive access on a developer's machine.
  • The trust is spent at install, not at use. This code ran when the editor loaded it. There was no later moment to notice something odd before it acted.

What to do

  • If you use Cursor and installed or updated that extension on 29 October: follow Cursor's notice, which includes a check script, remediation scripts for macOS and Windows, and a full list of indicators. It also advises rotating credentials reachable from affected machines.
  • Check by key, not just by label. The simplest check looks for the svc-telemetry comment. Cursor also publishes the key's fingerprint, and Phorion suggests checking "the public-key fingerprint rather than relying solely on the comment string", since a comment is easy to change.
  • Know where your editor gets its extensions. Different editors use different marketplaces, even for extensions with the same name. It's worth knowing which one yours uses.
  • Pin and review extension updates on important machines. Automatic updates mean a bad build reaches you without a decision on your part. On machines with access to production or client systems, consider updating deliberately.
  • Look at authorized_keys now and then. Any key you don't recognise in that file is a door someone else holds a key to.

Update: what Microsoft added

Microsoft's security team published its analysis on 30 October 2026, shortly before we first published this post, so it wasn't in our original sources. Four points from it are worth adding.

Which channels were clean

Microsoft says it has "confirmed that the malicious packages found on Cursor marketplace were never published nor distributed from Microsoft Visual Studio Marketplace and from npm @anthropic-ai/claude-code, which were not affected by this incident." It also reports that "Anthropic confirmed to Microsoft that their systems were not compromised and they did not develop the malicious package posted in Cursor marketplace." And: "At this time, all evidence suggests that the incident is limited to Cursor marketplace."

What to look for on disk

Cursor warned that uninstalling may not be enough. Microsoft says why: "Removing or updating the extension does not remove artifacts written outside its installation directory." It lists them. For the two platforms Cursor says were served:

  • macOS: a user LaunchAgent at ~/Library/LaunchAgents/com.apple.sysmond.plist (set with RunAtLoad and KeepAlive, so it starts and keeps running), plus the files ~/.config/.smt and ~/.config/.smtk.
  • Windows: an attempted scheduled task named SysMon, the files %USERPROFILE%\.config\smt.cmd and %USERPROFILE%\.config\smtk, and a fallback sysmon.cmd in the user's Startup folder.

Microsoft is careful about what a match means: "the presence of an artifact alone does not establish that the startup mechanism ran successfully", and "A result is an investigative lead, not proof of compromise." But it also explains why cleanup matters even where Cursor says SSH access was blocked: "the loader may still write keys and persistence artifacts even when tunnel access fails."

A name wasn't a guarantee at cleanup either

This post's lesson was that a publisher name doesn't prove a build is genuine. Microsoft makes the same point one step later: "The startup objects use names such as SysMon and com.apple.sysmond, which resemble monitoring components." These are the attacker's own files, given names that look like something you'd expect to find, which makes them easy to skip when you check what starts on your machine. Microsoft's advice: "Inspect each startup object's configured action before removal; names alone are not sufficient to distinguish malicious objects from legitimate monitoring software."

One more control

We suggested pinning and reviewing extension updates. Microsoft points to Cursor's own hardening guidance, which names two settings: "restricting extensions to approved publishers and configuring an extension installation cooldown to delay installation of newly published versions." A cooldown means a newly published version isn't installed straight away. That gap is what let this one reach machines while it was live.

Microsoft's analysis also agrees with the detail above that only the names of environment variables were sent: "The callback sends matching environment-variable names, not values."

Sources: Cursor, "Security notice — malicious version of Anthropic.claude-code (2.1.312 and 2.1.313) identified in the Cursor VS Code Extension Marketplace" (29 October 2026, updated 30 October; read 31 October 2026); Phorion, "Official Claude Code Extension Compromise: Security Advisory" (30 October 2026, updated 31 October; read 31 October 2026). Phorion sells endpoint detection and response software. Added 2 November 2026: Microsoft Security, "Inside a malicious Claude Code-branded extension and its persistent SSH backdoor" (published 30 October 2026; read 2 November 2026 via Microsoft's security blog feed).

#sidtechno#cybersecurity
All Articles