Does the cursor headless CLI create a codebase index for the analyzed codebase if only the cursor CLI is installed and not the IDE?

I want to use cursor CLI to make a code analysis tool, but I’m not sure if using the headless CLI will cursor build a codebase index for better code analysis results. thanks orz

Why no one from cursor team is replying to questions related to CLI and codebase indexing? Seems like a secret sauce.

Hi — it would be good to nudge this up.

Can you confirm the expected consistency of document/codebase indexing between the Cursor IDE and the standalone CLI? The public docs are clear that indexing exists when a project is opened in Cursor, but I cannot find a clear statement that a CLI-only installation creates or reuses the same semantic index.

Specifically, if only the CLI is installed, should we expect:

  1. full codebase/document indexing equivalent to the IDE,

  2. reuse of an index previously created by the IDE,

  3. or no persistent semantic index at all, with context limited grep/ripgrep-style search, to explicitly provided files and working directory state?

A understanding here matters for how we standardise agent workflows and what level of repo awareness we can rely on in headless/CLI usage.