SAIL CVE-2026-59959

The target build must include SAIL's built-in WebP codec (the advisory notes vulnerable builds are those compiled with that codec), and the application must accept and decode an attacker-supplied, malformed animated WebP whose container or ANMF frame metadata is inconsistent enough that WebPDemuxGetI returns a zero WEBP_FF_CANVAS_WIDTH; that zero width is then used as a divisor in sail_codec_load_init_v8_webp() before any generic image-skeleton validation in sail_advanced.c. Limiting factors: only SAIL versions <= 1.0.1 are affected (fixed in 1.0.2); builds without the WebP codec are unaffected; the CVSS vector requires passive user interaction (UI:P) - i.e. the victim application must process the file; and impact is confined to availability (a reliable single-process crash) with no confidentiality or integrity consequence and no code execution.

This is a medium-to-low real-world priority, not a high-urgency item. The only supplied score is CVSS 4.0 (4.8) with vector AV:L/AC:L/AT:N/PR:N/UI:P/VC:N/VI:N/VA:L, i.e. no confidentiality or integrity impact and only limited availability impact (a single-process crash), with a local attack vector and passive user interaction required. The impact ceiling is denial of service via process abort (CWE-369 divide-by-zero); there is no code execution or data exposure. No CISA KEV listing is present, so there is no confirmed active exploitation, and no EPSS probability was provided in the input, so the likelihood-of-exploitation signal is simply missing rather than low - that gap should be stated to decision-makers rather than assumed benign. A public PoC reference (a Python generator script plus full reproduction steps and ASAN FPE output) is included in the GitHub Security Advisory, which raises practical reproducibility even though the input does not formally mark exploit code as confirmed. The vendor/triage Decision label is 'this_month', consistent with a bug worth patching on a normal maintenance cadence rather than an emergency. One signal worth verifying: the supplied 4.0 vector is AV:L, but the description targets 'applications accepting untrusted WebP images' - in a network-facing upload/decoding service the realistic vector could be network-reachable with no per-instance user interaction, which would increase availability exposure; defenders of server deployments should treat the attack surface as remote-reachable even though the documented reproduction is a local CLI decode.

An attacker crafts a malformed animated WebP whose container/ANMF frame metadata is inconsistent, causing the WebP demuxer to report a zero canvas width, and delivers it to an application that decodes untrusted WebP input using SAIL (<= 1.0.1) built with the WebP codec. When the loader initializes, SAIL divides the computed bytes-per-line by the zero width, raising an arithmetic fault that aborts the process and denies service. A public PoC generator script and step-by-step reproduction (including ASAN FPE output) are published in the GHSA advisory.