Verified Advisories
Concise advisories for CRITICAL/HIGH vulnerabilities – affected and fixed versions confirmed against the vendor's own advisory before publishing. 681 total.
Path traversal in LXD backup tarball restore enabling root file write / RCE (CVE-2026-66898)
LXD's backup import endpoint (POST /1.0/instances with Content-Type: application/octet-stream) reads the instance name from the tarball's backup/index.yaml verbatim, with no path validation, and uses it as an on-disk ...
Fixed in 6.1; 5.21.2; 5.0.4; 4.0.12
Symlink following vulnerability enabling root RCE in LXD (CVE-2026-63294)
LXD fails to validate backup.yaml when it exists as a symbolic link during image or backup archive unpacking. An unprivileged, project-confined user with image and instance creation rights can supply a crafted archive...
Fixed in 4.0.12; 5.0.8; 5.21.6; 6.10
Symlink following arbitrary file read/write in LXD (CVE-2026-63293)
LXD fails to validate whether the metadata.yaml file within an image archive is a symbolic link during image import or unpacking. An authenticated user can supply a crafted image archive containing a symlinked metadat...
Fixed in 4.0.12; 5.0.8; 5.21.6; 6.10
Authorization bypass of project disk limits in LXD storage volumes (CVE-2026-63299)
LXD contains two code paths that allow an authenticated user to bypass project-level disk and volume quotas. The `storagePoolVolumeTypePostMove` function performs a cross-project volume move without invoking the `Allo...
Fixed in 5.0.8; 5.21.6; 6.10
Authorization bypass via instance migration config override in LXD (CVE-2026-63296)
An authorization bypass vulnerability in LXD allows an authenticated user in a restricted project to supply arbitrary configuration overrides during instance migration without any project restriction enforcement. The ...
Fixed in 5.0.8; 5.21.6; 6.10
Newline injection in LXD NVIDIA instance configuration handling (CVE-2026-63298)
LXD's `nvidia.driver.capabilities` instance configuration key uses the `validate.IsAny` validator, which accepts any input without filtering, allowing newline characters to pass through unsanitized into the generated ...
Fixed in 4.0.12; 5.0.8; 5.21.6
Authorization bypass via TOCTOU in LXD cross-project instance copy (CVE-2026-63297)
When copying an instance across projects in LXD, the target project restriction check runs before source instance configuration is merged into the request. This allows an authenticated attacker with view access to a s...
Fixed in 5.0.8; 5.21.6
Shell injection via unescaped parentheses in Shescape on Windows CMD (CVE-2026-73414)
Shescape, a JavaScript shell escape library, fails to escape parentheses (`(` and `)`) when using the `escape` or `escapeAll` APIs on Windows with the shell set to `cmd.exe` or `true` when CMD is the default. An attac...
Fixed in >=2.1.14 <3.0.0; >=3.0.1
Authenticated remote code execution in ScadaLTS (CVE-2026-19656)
ScadaLTS 2.7.8.1 exposes a server-side method without authorization checks, allowing any authenticated user, including those with only low-privilege read-only permissions, to execute arbitrary operating system command...
Fixed in 2.8.1; 2.8.2
Authorization bypass in LXD cross-project cluster instance migration (CVE-2026-62420)
An authorization bypass vulnerability in LXD allows an authenticated attacker with project-level access to move an instance cross-project to a different cluster member via POST /1.0/instances/{name}, causing the desti...
Fixed in 5.0.8; 5.21.6; 6.10
Security restriction bypass in LXD cross-project instance migration (CVE-2026-63300)
The `instancePostMigration` function in LXD fails to validate an instance's configuration against the target project's enforced restrictions when moving an instance between projects. An authenticated attacker with `ca...
Fixed in 5.0.8; 5.21.6; 6.10
Missing publish-password check in SiYuan graph API endpoints (CVE-2026-72804)
The `getGraph` and `getLocalGraph` endpoints in SiYuan filter document access by visibility tier only and never evaluate the publish password. Because password-protected documents are marked Visible = true, their node...
Fixed in v3.7.4
Publish-access bypass in SiYuan embedded block endpoints (CVE-2026-72795)
The `/api/block/getBlockDOMWithEmbed` and `/api/block/getBlockDOMsWithEmbed` endpoints in SiYuan apply publish-access checks only to the directly requested block, not to any blocks inlined by that block's embed (trans...
Fixed in v3.7.4
Session cookie signing key disclosure in siyuan via getConf endpoint (CVE-2026-72794)
The POST /api/system/getConf endpoint in siyuan returns the server's session cookie signing key (Conf.CookieKey) in its response body to unauthenticated users when publish mode is enabled with authentication disabled,...
Fixed in v3.7.4
Information disclosure in SiYuan publish-mode attribute view filter (CVE-2026-72798)
Two defects in SiYuan's `renderAttributeView` publish-access filter allow anonymous readers to retrieve restricted content in publish mode. First, Relation and Rollup cells carry mirrored content from related database...
Fixed in v3.7.4
Sensitive configuration disclosure in SiYuan via /api/system/getConf (CVE-2026-72793)
The `/api/system/getConf` endpoint in SiYuan is reachable by anonymous users and publish-reader tokens, and its non-administrator masking chain fails to suppress three fields that the configuration-export path explici...
Fixed in v3.7.4
Authorization bypass in SiYuan encrypted notebook publish access gate (CVE-2026-72789)
SiYuan's publish-access gate uses an opt-out list (`publishAccess.json`) that is never populated with encrypted notebooks, causing the gate to treat all encrypted notebooks as publicly accessible by default. While an ...
Fixed in v3.7.4
SQL injection in Budibase MySQL integration (CVE-2026-73300)
The MySQL integration in Budibase is configured with `multipleStatements: true`, allowing execution of multiple SQL statements in a single query. Attackers can inject malicious SQL commands through user input fields, ...
Fixed in 3.40.0
OS command injection in IBM Db2 Mirror for i (CVE-2026-16956)
IBM Db2 Mirror for i contains an OS command injection vulnerability (CWE-78) caused by improper neutralization of special elements in OS commands. A remote, unauthenticated attacker can exploit this flaw to execute ar...
Fixed in 7.4 PTF SJ10957; 7.5 PTF SJ10954
Server-side template injection in @prompty/core Nunjucks renderer (CVE-2026-73299)
The TypeScript Nunjucks renderer in @prompty/core evaluated untrusted .prompty template bodies with unrestricted JavaScript member access. An attacker-controlled template could traverse constructor and prototype prope...
Fixed in @prompty/core >= 0.1.5; @prompty/core >= 2.0.0-beta.5