DuneSlide: la inyección de prompts escapa del sandbox de Cursor hasta un RCE del sistema
Cato AI Labs (1 de julio de 2026) demuestra que un solo prompt manipulado — a través de un servidor MCP o un resultado web — puede sobrescribir el binario de sandbox de Cursor y lograr un RCE del sistema sin clic. Corregido en Cursor 3.0.
# DuneSlide: prompt injection escapes Cursor's sandbox (illustrative, defensive) # Zero-click indirect injection arrives via an MCP result or fetched web page: hidden = "[hidden instruction in untrusted content]" # It steers run_terminal_cmd's OPTIONAL working_directory outside the project, # silently widening the sandbox write allowlist a model-set arg controls: run_terminal_cmd(working_directory="[system path]", command="[REDACTED]") # -> next command runs unconfined -> OS-level RCE as the developer. # Defense: upgrade to Cursor 3.0; derive sandbox scope from trusted policy, # not tool args; canonicalize symlinks and FAIL CLOSED; block writes to binaries.