Critical Code Execution Vulnerability in IBM Langflow Allows Unauthenticated Attackers to Compromise Secrets and Establish Persistence
Overview
CVE-2026-10134 in IBM Langflow OSS exposes a critical flaw with a CVSS 10.0 score, enabling attackers to bypass authentication, exfiltrate secrets, and achieve persistent arbitrary code execution. This vulnerability poses immediate risks to organizations using affected versions for AI workflow orchestration.
Technical Analysis
The vulnerability stems from unsafe deserialization of user-provided `tool_code` in Langflow's API endpoints. By crafting malicious payloads in POST requests to `/api/v1/build/...`, attackers can read all process secrets, modify database contents, connect to internal services, and leverage cloud metadata endpoints. Attackers can persist by injecting malicious code into public flows, triggering automatic execution during legitimate user interactions.
Enterprise & DIB Impact
For Defense Industrial Base (DIB) and enterprise systems relying on Langflow for secure workflows, this flaw risks exposure of sensitive intellectual property, cloud infrastructure credentials, and tenant-isolation breaches. Attackers could disrupt operations by modifying critical AI workflows or escalating access across multi-tenant environments.
Recommended Actions
- Upgrade to Langflow 1.10.0 or later
- restrict API access using network segmentation and authentication headers
- validate and sanitize all user-submitted code payloads
- enable anomaly detection for unusual flow modification patterns
- and audit cloud metadata endpoint access controls
Need Help Assessing Your Exposure?
Axiom Cyber Research provides vulnerability assessment, red team operations, and security advisory services to the Defense Industrial Base and regulated sectors.
Schedule a ConsultationFull security advisory on 247alerts.net →