Is my AI tool safe?Coding agentsAnysphere
Is Cursor safe?
No tool is safe or unsafe on its own: it depends on what it can reach and how it is set up. The record for Cursor: 8 incidents in this center's library and 1 CVE in its advisory feed (1 critical), the newest published Aug 4, 2025. Below: what went wrong, what was fixed, and how to use it safely.
What to watch for
Most incidents share one shape: something the editor reads (a Slack message, a swapped MCP config, a support ticket, a rules file, a package's config) became instructions, and the agent ran them with the developer's permissions. On October 29, 2026 Cursor's own extension marketplace served a tampered Claude Code extension for about eleven hours that planted an SSH key, a reverse tunnel and startup items.
How to use it safely
Turn auto-run off for untrusted workspaces, review .cursor/ and rules files in code review, pin MCP servers to their tool lists, allow extensions only from approved publishers, and keep production credentials off developer machines.
What it can reach
- Reads attacker-controlled content
- Runs commands with your permissions
- Holds API keys and tokens
- Installs plugins, skills or tools from registries
What went wrong: incidents in the library
- A tampered Claude Code extension in Cursor's marketplace planted an attacker's SSH key, a reverse tunnel and persistence High
Cursor reported that modified copies of the Claude Code extension, versions 2.1.312 and 2.1.313, were available in its VS Code extension marketplace from about 2:27am to 1:35pm Pacific time on October 29, 2026, and that they… - LiteLLM 1.107.0: stolen GitHub tokens pushed a credential-stealing release to PyPI and Docker Critical
LiteLLM reported that on October 21, 2026 its release pipeline published litellm 1.107.0 with compiled malware that steals credentials and loads remote-code execution tools. Using three engineers' GitHub accounts, through stolen… - GitSpawn: a repository's .git/config runs code the moment seven coding agents open the folder High
Manifold Security (Francisco Rosales and Ax Sharma) showed on September 1, 2026 that AI coding agents run git commands such as git status and git diff when they open a folder, to gather context, before the workspace-trust prompt… - TrapDoor: packages steal secrets and plant hidden orders for AI coding assistants High
Socket found more than 34 malicious packages across npm, PyPI and crates.io, dressed up as wallet and smart-contract security tools. Besides stealing wallet keys, SSH keys and cloud and GitHub tokens, they wrote hidden… - CurXecute: a Slack message leads to code execution in Cursor High
A message in a public Slack channel could hijack Cursor's agent and make it run attacker commands. The agent could add an MCP server entry to its config, and Cursor started it before the developer approved anything. - MCPoison: a swapped MCP config gives persistent code execution in Cursor High
Cursor asked developers to approve an MCP server once and then trusted it forever, even if its command changed. A collaborator could commit a harmless config, wait for approval, then swap in a malicious command. - Supabase MCP: a support ticket tricks a Cursor agent into leaking secrets High
A support ticket with hidden instructions could make a developer's Cursor agent read a secret database table and post it back into the ticket for the attacker, because the agent connected with the all-powerful service_role key. - Rules File Backdoor: invisible instructions in AI coding rules files Medium
The rules files that tell AI coding assistants how to write code can carry invisible instructions. A poisoned rules file could make Cursor or Copilot quietly add malicious code, such as a script from an attacker's site, without…
Known CVEs
- CVE-2025-54135 Critical Cursor remote code execution via prompt-injected MCP config write
How many people use it
- $4B run rate Annualized revenue, as reported by Forbes (June 2026); Cursor does not publish user counts · Second Talent, compiling Forbes and company statements
A high CVE count can mean heavy scrutiny as much as weak engineering, so read the incidents and advisories, not only the number. The counts are what this center has recorded, not every flaw ever found; usage figures are the vendors' and registries' own published numbers.
Sources
- Second Talent, compiling Forbes and company statements: Annualized revenue, as reported by Forbes (June 2026); Cursor does not publish user counts
- Cursor: Claude Code extension security notice
- Cursor: An update on our recent security incident
- Microsoft Security Blog: Inside a malicious Claude Code-branded Cursor extension and its persistent SSH backdoor
- Help Net Security: Claude Code extension poisoned with persistent SSH backdoor
- LiteLLM: Oct 21, 2026 release v1.107.0 compromised
- Help Net Security: LiteLLM hit by second supply chain attack this year
- Wiz: LiteLLM compromised, steals credentials
- The Register: LiteLLM supply chain attack started with an exposed master key
- LiteLLM: the attack chain behind the v1.107.0 release
- Manifold Security: GitSpawn, AI coding agents hijacked through git config
- Goose advisory GHSA-r5pp-p5r8-466r (CVE-2026-72718)
- paddo.dev: Opening the folder was the exploit
- CoinDesk: wallet data targeted in the TrapDoor package attack
- Phoenix Security: TrapDoor cross-ecosystem campaign and AI assistant poisoning
- Aim Labs (Cato Networks): CurXecute
- Cursor advisory GHSA-4cxx-hrm3-49rm
- Check Point Research: MCPoison
- Cursor advisory GHSA-24mc-g4xr-4395
- General Analysis: Supabase MCP can leak your SQL database
- Supabase: defense in depth for MCP servers
- Simon Willison: the lethal trifecta in action
- Pillar Security: weaponizing code agents through rules files
- The Hacker News: Rules File Backdoor attack