CVE-2026-109375
6.1 MEDIUMpublic exploit availablePublished 2026-10-12 · Updated 2026-10-16
AI analysis for this CVE has not been generated yet. Raw NVD data is shown below.
NVD description
Vercel deepsec from 0.1.0 before 1.1.17 contains a link following vulnerability in extractTarballLocally that allows sandboxed code to write files outside the sandbox by planting links in result tarballs. Attackers controlling sandbox code, such as a prompt-injected coding agent, can plant symbolic or hard links in the host data directory to write arbitrary files on the orchestrator host.
CVSS vector
CVSS:3.1/AV:N/AC:H/PR:N/UI:R/S:C/C:N/I:H/A:N
Weaknesses
CWE-59
Public exploit & PoC references
All references
- https://github.com/vercel-labs/deepsec
- https://github.com/vercel-labs/deepsec/commit/c5a4e4220ece72aee67c2ea21f4340e65f673fc1
- https://github.com/vercel-labs/deepsec/pull/46
- https://github.com/vercel-labs/deepsec/pull/47
- https://www.vulncheck.com/advisories/vercel-deepsec-before-1.1.17-arbitrary-file-write-via-symlink-in-sandbox-tarball-extraction
Source data: NVD (nvd.nist.gov), public domain. Exploit-DB.ai adds local AI analysis for defensive use only.
Related CVEs
- HIGHCVE-2026-100419PoC
- HIGHCVE-2026-100690PoC
- HIGHCVE-2026-100692PoC
- CRITICALCVE-2026-100715PoC
- CRITICALCVE-2026-100716PoC
- HIGHCVE-2026-100838PoC
- CRITICALCVE-2026-101894PoC
- HIGHCVE-2026-102113PoC
Related by shared AI tags and CWE weakness class. Browse the full archive.