Koffei

CVE-2026-28797

🟠 Medium

InfiniFlow RAGFlow (open-source RAG engine) · up to 0.24.0

CVSS (Koffei note)
8.7
CVSS (v4.0, CVE List)
8.7
EPSS (2026-11-18)
0.583%
48.60th percentile

Exploitation

Not confirmed — listed in the ARTEX agents' vulnerability playbook for AI-credential theft (Amazon MadPot, 03/10 to 05/11/2026, Chinese-nexus groups, moderate confidence), but Amazon does not confirm a successful exploitation of this CVE. CISA-ADP SSVC poc (06/04/2026, older than the Amazon report); not in KEV. Rated 🟠: public PoC and no patch listed. cyberveille RSS checked 14/11/2026: no result

Timeline

  1. 03/04/2026CVE record published (CVE List)
  2. 13/11/2026Amazon: Chinese-nexus groups run cybercrime campaigns with ARTEX AI agents

Relationships

CVE-2025-3248CVE-2025-3248InfiniFlow (RAGFlow)InfiniFlow (RAGFlow)ARTEX AI credential-theft campaignsARTEX AI credential-theft…CVE-2026-28797CVE-2026-28797

exploits suspected named in the same story vendor · Select a box to open its page · All relationships

Risk

8.7 High (v4.0, GitHub CNA) / 8.8 High (v3.1, CISA-ADP), EPSS 0.583% (OpenCVE) (CWE-1336 / 94 / 78: unsandboxed Jinja2 template rendering lets an authenticated user run code, SSTI)

Fix

No — neither the CNA record nor OSV lists a fixed version (to verify)

Official description

RAGFlow is an open-source RAG (Retrieval-Augmented Generation) engine. In versions 0.24.0 and prior, a Server-Side Template Injection (SSTI) vulnerability exists in RAGFlow's Agent workflow Text Processing (StringTransform) and Message components. These components use Python's jinja2.Template (unsandboxed) to render user-supplied templates, allowing any authenticated user to execute arbitrary operating system commands on the server. At time of publication, there are no publicly available patches.

CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N

CWE-1336,CWE-20,CWE-78,CWE-94