CVE-2026-70149

In H3C Magic NX15 NX15V100R017, the write function of /api/esps in the RPC Interface allows manipulation of the path argument to lead to code injection. The attack can be executed remotely, and the exploit has been publicly disclosed.

In H3C Magic NX15 100R017, the function sub_416874 in /usr/bin/usrlist allows OS command injection. Remote exploitation is possible, and the exploit has been made public.

Pillow, a Python imaging library, from 2.6.0 until 13.0.0 is vulnerable to argument injection when invoking Ghostscript for EPS input. EpsImagePlugin.Ghostscript passes a caller-controlled input filename after -f without converting it to a safe absolute path, so a name beginning with a hyphen can be interpreted as a Ghostscript option.

Pillow prior to 13.0.0 has a JPEG decoder vulnerability where a full-width scanline can be written into a buffer sized for a requested draft scale. ImagingJpegDecode does not verify libjpeg's actual output width after scaling, and an 8-bit lossless SOF3 JPEG decoded by libjpeg-turbo 3.x can ignore the requested downscaling geometry.

Pillow prior to 13.0.0 has a FlashPix JPEG-tile decoder vulnerability that can write decoded scanlines beyond the allocated tile buffer. FlashPix (FPX) JPEG decoding selects raw mode LA with two bytes per pixel, while an unrecognized JPEG mode can be decoded as JCS_UNKNOWN with four components, causing ImagingJpegDecode to write four-component scanlines into the two-byte tile buffer.

Pillow has a heap buffer overflow and stale-heap disclosure in _tiffWriteProc when saving compressed TIFF data through libtiff to a target without a usable fileno(), such as BytesIO. The callback sizes its allocation from the prior buffer size instead of the current write position, allowing data to be copied beyond the allocation. This is fixed in version 13.0.0.

Grav API plugin (grav-plugin-api) 1.0.40 before 1.0.42 caches a full translation dictionary for every well-formed language code sent to the unauthenticated GET /api/v1/translations/{lang} endpoint, and 1.0.16 before 1.0.42 fails to check target and destination page rules in the sync and batch copy endpoints. Unauthenticated attackers can send small requests that each store about 560 KB to exhaust disk or APCu, while accounts with api.pages.write can overwrite translations or create pages that page-level rules deny.

Craft CMS from 5.0.0 before 5.11.0 contains an authorization bypass vulnerability that allows site-scoped GraphQL tokens to mutate other sites' content because requireAllowedSite() is not enforced. Attackers holding a site-A token can create entries in site B via supplied siteId and delete entries, categories, tags, and assets shared across sites.

Cognition before 0.13.0 contains an unauthenticated remote code execution vulnerability due to defaulting its agent sandbox backend to local execution and shipping no authentication on its REST/SSE API. Attackers can create a session via POST /sessions and post a message via POST /sessions/{id}/messages without credentials, causing the Deep Agents ReAct loop to run attacker-directed shell commands through the local shell backend, resulting in remote code execution as the server process.

Rank Math SEO PRO for WordPress is affected by a supply-chain compromise in a subset of version 3.0.123 packages downloaded or updated between October 15, 2026 20:02 UTC and October 16, 2026 07:12 UTC. The affected plugin files contained malware that could compromise a website if activated, establish persistent access, communicate with rank1cdn[.]com, and expose WordPress credentials, database credentials, API keys, and other secrets.