CVE-2025-55346 — AI Deep Analysis Summary

🚨 **Essence**: Flowise AI suffers from a critical **JS Injection** flaw. User inputs feed directly into unsafe dynamic function constructors.…

🛡️ **Root Cause**: **CWE-94** (Code Injection). The vulnerability stems from improper neutralization of special elements used in a code command.…

📦 **Affected**: **Flowise** (the open-source tool by FlowiseAI for building LLM apps). The data does not specify exact version numbers, but any instance running this vulnerable component is at risk.…

👑 **Privileges**: **Full Host Control**. The CVSS score is **Critical (9.8)**. Attackers gain High Confidentiality, Integrity, and Availability impact.…

⚡ **Threshold**: **LOW**. CVSS Vector: `AV:N/AC:L/PR:N/UI:N`. - **Network**: Remote exploit. - **Complexity**: Low. - **Privileges Required**: None. - **User Interaction**: None.…

📢 **Public Exp**: **Yes**. A third-party advisory from **JFrog Research** (JFSA-2025-001379925) confirms the vulnerability.…

🔍 **Self-Check**: 1. Check if you are running **Flowise**. 2. Look for dynamic function usage in custom nodes or workflows. 3. Scan for **JS Injection** patterns in input fields. 4.…

🩹 **Official Fix**: The CVE was published on **2025-08-14**. The provided data does not list a specific patch version or mitigation guide yet.…

🚧 **Workaround**: 1. **Isolate**: Run Flowise in a strict container (Docker/K8s) with minimal privileges. 2. **Input Sanitization**: Validate and escape all user inputs before they reach function constructors. 3.…

🚨 **Urgency**: **CRITICAL**. With a **CVSS 9.8** score and **Remote Code Execution** potential without authentication, this is a **P0** incident. Immediate patching or isolation is required.…