GitHub Enterprise Server EUVD-2026-28464

Exploitation requires network access to the GitHub Enterprise Server instance and the ability to trigger notebook viewing functionality. The attacker must be able to submit or reference a notebook file containing malicious URLs, either by creating repository content, submitting pull requests, or referencing external notebook URLs depending on the specific notebook viewer implementation and access controls. The vulnerable instance must have network connectivity to the targeted internal services for the SSRF to successfully retrieve data. No authentication is required per the CVSS vector (PR:N), though practical exploitation may be limited by repository access policies if notebook content must be hosted within the GitHub instance rather than referenced externally. Attack success depends on the presence of accessible internal services that respond to HTTP requests and return sensitive information without additional authentication.

This vulnerability presents significant real-world risk for organizations running GitHub Enterprise Server in environments with sensitive internal services. The CVSS 7.9 score appropriately reflects the attack characteristics: network-accessible (AV:N), low complexity (AC:L), no authentication required (PR:N), and high impact to subsequent systems. The absence from CISA KEV and no public exploit code identified suggests no widespread active exploitation at time of analysis, though the bug bounty report indicates at least proof-of-concept capability exists. SSRF vulnerabilities are inherently valuable to attackers as they enable internal network reconnaissance, access to metadata services (cloud instance credentials), internal API exploitation, and potential data exfiltration from services not directly internet-facing. The attack automation potential is moderate - while exploitation is straightforward once a target is identified, successful attacks require knowledge of internal network topology and service endpoints. Priority depends heavily on deployment context: critical for instances with access to cloud metadata endpoints, internal APIs, or sensitive backend services; lower risk if deployed in isolated network segments with strict egress filtering already implemented.

An external attacker identifies a GitHub Enterprise Server instance and crafts a malicious notebook containing specially encoded URLs designed to exploit parser confusion. When the notebook viewer processes this content, the validation layer interprets the URL as pointing to an approved external resource, but the HTTP request library parses it to target an internal service such as the cloud metadata endpoint (169.254.169.254) or internal API server. The attacker retrieves sensitive data such as instance credentials, internal API responses, or configuration information, which can be exfiltrated through the notebook viewer's response or used to pivot deeper into the organization's infrastructure.