PonoLens guide

What does Cursor send when you ask it a question?

Cursor needs context to answer coding questions. That context can include more than the words you type.

The short answer

When you use an AI feature, Cursor may send your prompt and relevant code context to Cursor and model providers so they can produce an answer. The exact information depends on the feature, your settings, and the files Cursor chooses as context.

Code context can include open files, selected code, file names, and other project information that helps answer the question. Cloud features may handle code differently from an ordinary question.

What PonoLens can show

When Cursor provides a supported hook or activity record, PonoLens can connect the prompt with reported commands and file changes. It can also warn when the available text contains possible private information.

Coverage has limits. PonoLens can only show information Cursor makes available through a supported connection. It does not inspect encrypted internet traffic or promise a complete copy of everything Cursor sends.

Before you ask

  • Remove passwords, API keys, access tokens, and customer details.
  • Share only the files needed for the task.
  • Check Cursor’s current privacy and data settings.
  • Review the PonoLens activity card after the prompt.

For Cursor’s own description of its data flows, read its Privacy and Data Governance documentation.

Read next

How to check which files an AI coding agent changed →