Medium severity · published Mar 11, 2026
CVE-2026-32128
FastGPT Python Sandbox Bypass of File-Write Restriction
6.3
CVSS
0.3%
EPSS
No
On KEV
Description
FastGPT is an AI Agent building platform. In 4.14.7 and earlier, FastGPT's Python Sandbox (fastgpt-sandbox) includes guardrails intended to prevent file writes (static detection + seccomp). These guardrails are bypassable by remapping stdout (fd 1) to an arbitrary writable file descriptor using fcntl. After remapping, writing via sys.stdout.write() still satisfies the seccomp rule write(fd==1), enabling arbitrary file creation/overwrite inside the sandbox container despite the intended no file writes restriction.
CVSS vector
CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:L/I:L/A:L
Attack vector
Network
Complexity
Low
Privileges
Low
User interaction
None
Confidentiality
Low
Integrity
Low
Availability
Low
Scope
Unchanged
Exploitation
Is it being used?
Exploit prediction (EPSS)0.3% · higher than 24% of CVEs
Exploited in the wildNo evidence yet
CISA KEVNot listed
Ransomware campaign useNot known
Affected products
Which versions
FA
FastGPT · labring
<= 4.14.7 · source: vendor, NVD
References
NVD
https://nvd.nist.gov/vuln/detail/CVE-2026-32128
CVE.org
https://www.cve.org/CVERecord?id=CVE-2026-32128
Exploit
https://github.com/labring/FastGPT/security/advisories/GHSA-6hw6-mxrm-v6wj
More from FastGPT
10.0
CVE-2026-34162
FastGPT: Unauthenticated SSRF via httpTools Endpoint Leads to Internal API Key Theft
9.9
CVE-2025-49131
FastGPT Sandbox Vulnerable to Sandbox Bypass
9.8
CVE-2026-42302
FastGPT: Unauthenticated Remote Code Execution (RCE) via code-server Misconfiguration in agent-sandbox
9.8
CVE-2026-40351
FastGPT: NoSQL Injection in loginByPassword leads to Authentication Bypass
9.3
CVE-2026-68929
FastGPT: Unauthenticated WeChat channel hijack and denial of service via shareId-only authorization