Skip to content
Submit incident
AII-1000-74433
ReportedReported by fewer than three independent publishers, with no official finding.Supporting sources are pending verification.

Cursor AI Extension Vulnerabilities Expose Developer Credentials and Enable Code Execution

Researchers discovered high-severity vulnerabilities in the Cursor AI coding environment that allow malicious extensions to access stored developer credentials and execute arbitrary code on developers' machines. Cursor's use of unprotected local storage for API keys and session tokens without access controls enables attackers to extract sensitive data and compromise developer environments.

February 2026 Β· Cursor Β· Security and cyberattacks
February 2026
AII-1000-74433
Quick facts
Category
Security and cyberattacks
Date
February 2026
Drafted by
Raidu Incident Detection

What happened

Cursor stores authentication data including API keys and session tokens in a local SQLite database without protection from operating system keychains or access controls, according to LayerX research. Any installed extension can directly query this database regardless of the permissions it requests, allowing extraction of credentials without user interaction or visible alerts.

Researchers demonstrated that a malicious extension disguised as a harmless tool could gain code execution within Cursor, immediately access local credential storage, and silently exfiltrate sensitive data. Cursor reportedly acknowledged the notice but stated that defining trust boundaries is the user's responsibility. The issue remained unresolved as of April 28, 2026.

Reported impact

Stolen API keys can be used to access third-party platforms such as OpenAI, Anthropic or Google services, creating downstream risks including unauthorized API usage leading to financial loss, exposure of prompts and outputs, and potential misuse of services for further attacks.

LayerX assigned the flaw a CVSS score of 8.2, warning that it could enable full credential compromise across a developer's environment.

Index analysis

The vulnerability stems from Cursor's design choice to store secrets locally without standard security mechanisms and without enforcing access controls between extensions and storage. The absence of permission prompts or warnings increases the risk to developers who install extensions from marketplaces.

Classification

Category
Security and cyberattacks
Industry
Digital security
Harm status
Potential harm
Intent
Unintentional
Timing
Post-deployment
Harmed
Business
Failure modes
Data leak
AI systems
Cursor

Sources and evidence

Sources not yet published

This record was researched and written by the Index from original reporting. Its sources are still being verified by an editor, and they are listed here once that check is complete. Until then the record carries no reviewer.

How sources are verifiedSuggest a source for this incident