Home / Vulnerability Intelligence / CVE-2026-33075

CVE-2026-33075 - Vulnerability Analysis

HighCVSS: 8.8

Last Updated: March 23, 2026

FastGPT - Command Injection & Information Disclosure

Published: March 20, 2026Updated: March 23, 2026PoC AvailableRemote Exploitable

Overview

FastGPT <= 4.14.8.3 contains a command injection caused by use of pull_request_target workflow with attacker-controlled Dockerfiles in fastgpt-preview-image.yml, letting external contributors execute arbitrary code and exfiltrate secrets, exploit requires external contributor pull request.

Severity & Score

Severity: High

CVSS Score: 8.8

EPSS Score: 2.8%(Probability of exploitation in next 30 days)

Impact

External contributors can execute arbitrary code and exfiltrate secrets, enabling supply chain attacks via the container registry.

Mitigation

Update to the latest version once a patch is available.

References

https://github.com/labring/FastGPT/security/advisories/GHSA-xfx8-w35j-485c

Social Media Activity(1 post)

Offensive Sequence

@offseq

Mar 20, 2026

🚨 CRITICAL: CVE-2026-33075 affects labring FastGPT ≤4.14.8.3. GitHub Actions workflow flaw enables attackers to run code & steal secrets, risking supply chain compromise. No patch — audit workflows & restrict secrets now! https://radar.offseq.com/threat/cve-2026-33075-cwe-494-download-of-code-without-in-52a1ff21 #OffSeq #Infosec #SupplyChain

View original post

Related Resources

Details

CVE ID: CVE-2026-33075
Severity: High
CVSS Score: 8.8
Type: command_injection
Status: confirmed
EPSS: 2.8%
Social Posts: 1

CWE

CWE-494

CVSS Metrics

CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H

EPSS Score

2.8%Probability of exploitation in the next 30 days