A public Sentry key is all it takes to hijack Claude Code, Cursor, and Codex
On June 17, the Threat Labs team at Tenet Security, an AI-agent security startup newly out of stealth, documented a security vulnerability. The article's title indicates that a public Sentry key is sufficient to hijack AI-powered tools such as Claude Code, Cursor, and Codex. This method is implicitly referred to as "agentjacking" and an "Sentry MCP attack" in the article's title and URL. The documentation by Tenet Security's Threat Labs team highlights a security concern where access to a public Sentry key could lead to the compromise of these specific AI agent systems. The findings were published by The New Stack.
Developers using Claude Code, Cursor, and Codex face a security risk, as a public Sentry key can reportedly hijack these AI agent systems.