
Cursor has confirmed a security breach that allowed attackers to distribute a malicious Claude Code...
Cursor has confirmed a security breach that allowed attackers to distribute a malicious Claude Code extension through its marketplace. According to BleepingComputer, the attackers got in using credentials stolen from Fireworks AI, an AI infrastructure provider. The public details are still limited, but the incident shows how fragile the trust is around developer tooling and AI supply chains.
BleepingComputer reports that Cursor confirmed the breach itself. The attackers stole credentials from Fireworks AI and then used them to publish and distribute a malicious extension for Claude Code via the Cursor marketplace.
The sources available so far do not say:
I am not going to guess at any of these. The primary report and Cursor's own statements are the places to look for updates.
The story is a chain of trust failures across several parties:
For engineers, the last link is the one that hurts. Editor extensions usually run with the permissions of the logged-in user. A development machine commonly has access to source repositories, environment variables, SSH keys, cloud credentials and API tokens. An extension from a marketplace you trust looks legitimate, so the usual caution tends to be absent.
It also shows that AI tooling adds new dependencies to your threat model. Teams now rely on model hosts, inference providers, coding assistants and plugin ecosystems. Each one holds credentials and has some level of access, and a compromise at one can spread to the others.
The sources do not include specific mitigation steps, so this is general hygiene, not vendor-confirmed guidance.
This incident is still developing, and the facts above may change as more information becomes available. I will treat anything beyond the confirmed points as unknown until the primary sources say otherwise.
aiThe Dynatrace Cursor plugin brings an MCP server and around thirty skills to your editor to investigate production, and shows how to roll it out to a team.
mcpSeven web search APIs ship an official MCP server you can add to Claude, Claude Code or Cursor:...
aiAgent commands failing with DNS, 403 or read-only errors? Allow the right hosts, cache folders and localhost in Claude Code, Cursor and Codex sandboxes.
securityIf you use Cursor and had the Claude Code extension installed on 29 October 2026, there is a real...
cursorOn November 13, Cursor published the full investigation into a security incident it first disclosed...
How aggressive context compression in Cursor taught me to verify what my AI actually sees—and why mksglu/context-mode became non-negotiable in my workflow.
Workflows from the Neura Market marketplace related to this Cursor resource