LibreSSL CVE-2026-118371

The target application must use LibreSSL's libcrypto OCSP client path (OCSP_sendreq_new() / OCSP_parse_url()) and have OCSP revocation checking enabled, and the attacker must be able to present a certificate whose AIA OCSP URI they control and embed CR/LF characters into it. The AT:P metric reflects that the victim must actually initiate the OCSP request for the crafted URI before injection occurs. Limiting factors: exploitation yields only HTTP header injection / request splitting into OCSP traffic toward the responder or proxy (integrity-low, no confidentiality or availability impact), it requires OCSP checking to be active and to reach the attacker-influenced URI, and it affects only LibreSSL builds through 4.4.0 - versions incorporating commit d3f7195f, which rejects non-alphanumeric/non-punctuation and non-ASCII path characters, are not affected.

This is a moderate-severity integrity issue, not a high-urgency emergency. The supplied CVSS 4.0 score is 6.3 with vector AV:N/AC:L/AT:P/PR:N/UI:N and impact limited to VI:L (low integrity) while VC/VA and all subsequent-system metrics are N - meaning no confidentiality or availability impact and no cross-system scope change claimed. The AT:P (Attack Requirements: Present) metric is significant: exploitation is not fire-and-forget over the network but depends on the victim application actually performing an OCSP request against a certificate whose Authority Information Access URI the attacker controls. The decision label is 'this_month' (not act_now/emergency), which is consistent with these signals. There is no CISA KEV listing (not actively exploited) and no EPSS score is provided in the input, so probabilistic exploitation likelihood cannot be quantified here - state this gap explicitly rather than assuming low. No public exploit code or PoC reference is present among the references (they are the patch commit, source blobs, release notes and the VulnCheck advisory). Net assessment: a credible, vendor-confirmed CRLF/HTTP-request-injection flaw in a TLS library, but the real-world priority is capped by low technical impact (injection into OCSP traffic only), the AT:P precondition, and the absence of KEV/PoC - patch on the vendor's this_month cadence rather than treating it as a drop-everything issue.

An attacker crafts or obtains a certificate whose Authority Information Access OCSP URI embeds carriage-return/line-feed bytes, then induces a client or server running LibreSSL (through 4.4.0) to perform OCSP revocation checking on that certificate. Because OCSP_sendreq_new()/OCSP_parse_url() write the unvalidated path straight into the OCSP POST request line, the embedded CRLF lets the attacker inject extra HTTP headers or split the request sent to the OCSP responder or an intervening proxy. No public exploit identified at time of analysis.