The vulnerability exists in the oembed endpoint's URL processing chain. The primary vulnerable function is fetchOembedData which handled user-provided URLs and made network requests without adequate validation of target hosts (IP/localhost checks added in patch). The removed getOembedUrlFromHTML function contributed to the vulnerability by extracting URLs from untrusted HTML content that were then used in requests. The patch adds multiple security layers: 1) isIpOrLocalhost validation for initial URLs and embedded oembed links 2) Response validation against oembed spec 3) Removal of direct HTML parsing via getOembedUrlFromHTML. During exploitation, the call stack would show fetchOembedData processing malicious URLs either directly provided or extracted from HTML content.