Wazuh Shows How to Detect Exploitation of Langflow CVE-2026-33017
Summary
A Wazuh walkthrough uses nginx access logs, auditd, and custom detection rules to identify attempts to exploit Langflow’s unauthenticated code-injection flaw, CVE-2026-33017, and resulting host activity.
Key points
- CVE-2026-33017 allows unauthenticated remote code execution through Langflow’s public flow build endpoint in versions before 1.9.0.
- The flaw stems from accepting attacker-controlled flow data that can include Python code evaluated during workflow processing.
- The article configures nginx to log requests to the vulnerable endpoint and auditd to monitor commands executed by the Langflow process.
- Custom Wazuh decoders and rules flag endpoint requests, inline Python components, execution primitives, and reconnaissance commands.
- The article reports active exploitation, including deployment of cryptocurrency-mining software and searches for environment variables and sensitive files.
- Upgrade to Langflow 1.9.0 or later; interim measures include restricting network access, avoiding public flows on exposed deployments, and running Langflow as an unprivileged user.
Article Details
- Defense Focus
- Detect exploitation of CVE-2026-33017 against Langflow by correlating requests to the public flow build endpoint with commands executed by the Langflow process.
- Detection Methods
- Parse nginx access-log entries and alert on requests to /api/v1/build_public_tmp/.
- Inspect request bodies for an inline Python component and code-execution primitives such as exec(), subprocess, socket, and system().
- Monitor auditd execve events for the Langflow service account and alert on reconnaissance commands; generate a higher-level alert when multiple such commands occur within 10 seconds.
- Correlate application-level requests with operating-system command activity to investigate exploitation attempts and resulting execution.
- Data Sources
- nginx reverse-proxy access log, including request body
- auditd log with execve events filtered by the Langflow effective user ID
- Rule Types
- Custom Wazuh decoders and rules
- PCRE2-based log parsing and field matching
- Frequency/timeframe correlation rule
- Platforms
- Wazuh
- Defensive Actions
- Upgrade Langflow to version 1.9.0 or later.
- Until upgraded, restrict network access to trusted sources and avoid making flows public on internet-facing deployments.
- Run Langflow as a dedicated, unprivileged service account rather than as root.
- Forward nginx access logs and auditd logs to Wazuh for monitoring.
MITRE ATT&CK
T1059.004 · Unix ShellThe article maps commands executed by the Langflow service, including reconnaissance commands detected through auditd, to this sub-technique.T1059.006 · PythonThe article maps execution of attacker-supplied Python code in the Langflow process to this sub-technique.T1190 · Exploit Public-Facing ApplicationThe article maps exploitation of Langflow's public flow build endpoint to this technique.
CVE
CVE-2025-3248CVE-2026-33017 follows a similar pattern to CVE-2025-3248, another Langflow remote code execution vulnerability.CVE-2026-33017CVE-2026-33017 is a critical code injection vulnerability in Langflow’s public flow build endpoint that allows unauthenticated remote code execution (RCE).CVE-2026-33309Langflow versions up to 1.8.1 also carry CVE-2026-33309, a separate path traversal and arbitrary file write vulnerability in POST /api/v2/files/, which is fixed in the 1.9.0 release.
Vendors
Products
Kali Linux A Kali Linux endpoint to exploit CVE-2026-33017 on an Ubuntu endpoint.LangflowLangflow is an open source, low-code platform for building and deploying AI-powered agents and workflows.NGINXWe place nginx in front of Langflow as a reverse proxy to record requests to the vulnerable build_public_tmp endpoint.UbuntuAn Ubuntu 24.04 endpoint with Wazuh agent 4.14.7 installed and enrolled in the Wazuh server.