CVE-2025-10854
The txtai framework allows the loading of compressed tar files as embedding indices. While the validate function is intended to prevent path traversal vulnerabilities by ensuring safe filenames, it does not account for symbolic links within the tar file. An attacker is able to write a file anywhere in the filesystem when txtai is used to load untrusted embedding indices
Scoring
- Severity
- HIGH
- CVSS base score
- 8.1
- CVSS vector
- CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:H/I:H/A:H
- EPSS probability
- 0.46%
- CWE
- CWE-61
- Published
- 2025-09-22
- Last modified
- 2026-03-12
Weakness type
Related vulnerabilities
- CVE-2026-57825 — In the opam package before 2.5.2 for OCaml, the sandbox protection mechanism can be bypassed...
- CVE-2026-79939 — Dell PowerProtect Cyber Recovery, versions Prior to 20.3, contain an UNIX Symbolic Link (Symlink)...
- CVE-2026-75038 — Predictable temporary file in /tmp allows symlink attack in LACT
- CVE-2026-55168 — Runtipi: Authenticated arbitrary file write via backup restore symlink planting
- CVE-2026-63125 — Incus vulnerable to root RCE via image backup.yaml symlink
- CVE-2026-64846 — Nix: Arbitrary file truncation outside the sandbox with recursive-nix experimental feature
- CVE-2026-32657 — Dell AppSync Version 4.6.0.0, Dell Metro Node Version 8.0.0, Dell UCC Edge Version 3.0.1, Dell...
- CVE-2026-47766 — crun follows rootfs /dev symlink while creating default devices