A security incident involving the Cursor IDE's extension marketplace has been detailed, stemming from a stolen Fireworks AI API key. This key exposed unredacted credentials within coding agent traces, which attackers then used to inject malicious code into two versions of the Claude Code extension. These compromised versions were available on the Cursor marketplace for approximately 11 hours, potentially affecting up to 5,000 users before being removed. The incident highlights a significant vulnerability where sensitive information in agent traces can lead to supply chain attacks. AI
IMPACT Highlights a critical security flaw in AI agent trace logging, emphasizing the need for robust redaction to prevent supply chain attacks.
RANK_REASON The article details a security vulnerability and a proposed solution, but does not announce a new frontier model or significant industry-wide event.
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →