Prompt injection remote code execution in FlowiseAI Flowise CSV Agent (CVE-2026-41264)

A missing sandbox boundary in the `run` method of the `CSV_Agents` class allows the LLM-generated Python script to be evaluated without proper isolation. An unauthenticated remote attacker can use prompt injection to cause the LLM to produce a malicious Python script that bypasses input validation and executes arbitrary OS commands in the context of the server process. Public POC available. Affected versions: - npm flowise <= 3.0.13 - npm flowise-components <= 3.0.13 We recommend updating to: 3.1.0. CVSS 9.2 · our assessment: act_now Reference: https://github.com/FlowiseAI/Flowise/security/advisories/GHSA-3hjv-c53m-58jj Details: https://vuln.today/cve/CVE-2026-41264