CVE-2025-61913
Published: 08 October 2025
Description
Adversaries may backdoor web servers with web shells to establish persistent access to systems.
Security Summary
CVE-2025-61913 is a path traversal vulnerability (CWE-22) affecting the WriteFileTool and ReadFileTool components in Flowise, an open-source drag-and-drop user interface for building customized large language model (LLM) flows. Versions of Flowise prior to 3.0.8 fail to restrict file path access in these tools, enabling unauthorized file operations. The vulnerability carries a CVSS v3.1 base score of 9.9 (AV:N/AC:L/PR:L/UI:N/S:C/C:H/I:H/A:H), indicating critical severity due to its potential for high-impact exploitation.
Authenticated attackers with low privileges can exploit this vulnerability remotely over the network with low complexity and no user interaction required. Successful exploitation allows reading and writing arbitrary files to any path on the file system, which can lead to remote command execution by targeting critical system files or scripts.
Flowise addressed this issue in version 3.0.8, as detailed in the project's security advisories (GHSA-j44m-5v8f-gc9c and GHSA-jv9m-vf54-chjj) and the fixing commit (1fb12cd93143592a18995f63b781d25b354d48a3). Security practitioners should update to Flowise 3.0.8 or later to mitigate the risk, per the release notes.
This vulnerability is particularly relevant in AI/ML environments, as Flowise is designed for LLM workflow orchestration, potentially exposing deployments handling sensitive model data or configurations to filesystem compromise. No public evidence of real-world exploitation has been reported as of the CVE publication on 2025-10-08.
Details
- CWE(s)
Affected Products
AI Security Analysis
- AI Category
- AI Agent Protocols and Integrations
- Risk Domain
- Other ATLAS/OWASP Terms
- OWASP Top 10 for LLMs 2025
- None mapped
- MITRE ATLAS Techniques
- None mapped
- Classification Reason
- Flowise is a drag-and-drop UI platform for building customized large language model (LLM) flows, fitting as an 'Other Platforms' category for AI workflow and orchestration tools.
MITRE ATT&CK Enterprise Techniques
Why these techniques?
Arbitrary file read enables T1005 (Data from Local System). The public-facing web app vulnerability maps to T1190 (Exploit Public-Facing Application). Arbitrary file write facilitates T1505.003 (Web Shell) for RCE and persistence.