Ghost CMS CVE-2026-70595

Exploitation requires a Ghost instance running v6.26.0 through v6.54.0 with at least one outbound-HTTP-making feature active (Webmentions confirmed; other features may also be vulnerable per the advisory). The CVSS vector PR:N and UI:N confirm exploitation requires neither authentication nor user interaction - any network-reachable Ghost endpoint processing such features is potentially exploitable. The AC:H rating indicates non-trivial exploitation complexity, though the specific technical barrier is not enumerated in available data and is inferred from the CVSS vector alone. Exploitation is concretely limited by the blind SSRF constraint: response content is never returned to the attacker, so only side-channel techniques yield useful intelligence. Deployments with no co-located sensitive internal services, or with egress filtering already blocking RFC 1918 destinations, face negligible practical risk despite the network-accessible attack vector.

The vendor-assigned CVSS 3.1 score of 4.0 (Medium) reflects network reachability (AV:N) and zero authentication/interaction requirements (PR:N, UI:N), but is moderated by high attack complexity (AC:H) and limited confidentiality impact (C:L) with no integrity or availability impact. The S:C metric elevates strategic risk by indicating the internal network - not just Ghost - is within the attack's reach. The blind SSRF nature (no response data returned) substantially limits attacker utility compared to full SSRF: exploitation requires side-channel inference (timing, error differentiation) rather than direct data extraction, which partially justifies AC:H. No EPSS score was provided in the input; exploitation probability cannot be quantified from available data. There is no CISA KEV listing and no known public POC, indicating this is not under active widespread exploitation at time of analysis. Real-world risk is highest for self-hosted Ghost deployments co-located on networks with sensitive unauthenticated internal services (cloud metadata APIs, internal dashboards, private databases); risk is substantially reduced in environments with strict egress filtering or network segmentation isolating the Ghost host.

An unauthenticated attacker sends a crafted Webmention HTTP request to a vulnerable Ghost instance, specifying a URL targeting an internal service - for example, a cloud instance metadata endpoint (169.254.169.254) or an unauthenticated internal API. Ghost's insufficient URL validation permits the outbound request, causing the server to probe the internal host on the attacker's behalf. Because no response data is returned, the attacker must rely on timing differences or HTTP error behavior to infer whether internal hosts are live, gradually mapping the internal network topology. No public POC has been identified at time of analysis.