When following redirects via `followRedirect`, the software reuses caller-supplied options across different origins. Specifically, the `#requestInternal()` function recursively calls itself while preserving the original `options` object. This causes sensitive headers and authentication values to be sent to the redirect target even if it has a different scheme, host, or port.
Credential-bearing headers that may be exposed include `Authorization`, `Cookie`, `Proxy-Authorization`, `x-api-key`, `x-auth-token`, and `x-access-token`. If a request is redirected to an attacker-controlled origin, these credentials can be captured and potentially reused against the original API or related services. This issue requires no user interaction.