CVE-2026-107329

Langflow code-generation input can cross an execution boundary

IBM / Langflow · Langflow OSS

First documented disclosure · Review checkpoint

Summary

Server-side code execution under application permissions.

Severity and exploitation evidence

Critical · 9.9/10 · CVSS 3.1. Severity source.

Exploitation not established: The reviewed primary records do not establish malicious exploitation or a confirmed victim.

Exploitation evidence source

Technical severity is not a measure of observed attacks or of risk to your deployment. Attempts, demonstrations, and successful compromise retain the qualifications in the source.

How the weakness works

Generated code incorporates insufficiently controlled input; exact patterns and endpoint details are undisclosed.

Conditions required

A remote authenticated account can reach the affected generation path.

Potential impact

Server-side code execution under application permissions.

Affected versions and configurations

IBM lists Langflow OSS 1.0.0 through 1.12.4.

Fix and mitigation status

1.12.5 per IBM; this is an issue-specific fix, not assurance that the old release addresses later findings.

Practical next steps

Upgrade to a current supported release containing the fix; back up the deployment and validate workflow behavior. Where MCP stdio is used, IBM additionally requires interpreter hardening and administrator-only controls; configuration changes do not replace upgrading.

Operating lesson

Authorize and isolate generated-code execution independently of text generation.

Evidence and limits

IBM bulletin 7293016 was initially published 2026-10-19; the CNA record followed on 2026-10-29. Earlier upstream patches are not assigned to this CVE by inference. Exact technical details absent from the public record remain unknown; a CVSS score is not an observed incident.

Sources