Skip to main content
Vulnerability Database/CVE-2026-97655

CVE-2026-97655: IBM Langflow OSS RCE Vulnerability

CVE-2026-97655 is a remote code execution vulnerability in IBM Langflow OSS versions 1.0.0 through 1.12.2 caused by an incomplete blocklist in the code security scanner. This article covers technical details, impact analysis, and mitigation strategies.

Published:

Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.

Sources: National Vulnerability Database, CISA KEV, Exploit-DB

CVE-2026-97655 Overview

CVE-2026-97655 is a code execution vulnerability in IBM Langflow OSS versions 1.0.0 through 1.12.2. The flaw stems from an incomplete blocklist in the product's code security scanner, which fails to catch all dangerous code constructs during validation. A remote authenticated attacker can bypass the scanner and execute arbitrary code on the host running Langflow.

Langflow is a widely used open-source framework for building large language model (LLM) applications and agent workflows. Exploitation grants attackers the ability to run commands with the privileges of the Langflow process, which typically has access to API keys, model credentials, and backend data sources.

Critical Impact

Authenticated remote attackers can execute arbitrary code on Langflow servers, compromising confidentiality, integrity, and availability of AI pipelines and connected systems.

Affected Products

  • IBM Langflow OSS 1.0.0 through 1.12.2
  • Deployments embedding Langflow as a component for LLM orchestration
  • Hosted environments running vulnerable Langflow container images

Discovery Timeline

  • 2026-10-07 - CVE-2026-97655 published to NVD
  • 2026-10-07 - Last updated in NVD database

Technical Details for CVE-2026-97655

Vulnerability Analysis

The vulnerability is classified as [CWE-94] Improper Control of Generation of Code (Code Injection). Langflow accepts user-supplied Python code for custom components and flow logic. Before execution, the platform's security scanner inspects submitted code against a blocklist of dangerous functions, imports, and attributes.

The blocklist is incomplete. Attackers can craft Python payloads that reach dangerous runtime capabilities through unlisted module paths, attribute lookups, or dynamic imports. Once the scanner approves the code, Langflow executes it inside the server process, giving the attacker arbitrary code execution.

The attack requires network reachability to the Langflow API and valid low-privilege credentials. User interaction is not required. Successful exploitation yields full control over the Langflow runtime and any secrets, model endpoints, or data stores reachable from it.

Root Cause

The root cause is reliance on a denylist filter to constrain untrusted code execution. Blocklists for Python sandboxing are structurally fragile because the language exposes many indirect paths to the same primitives, such as __builtins__, __import__, getattr, and object introspection via __class__ and __subclasses__. Any missed entry in the blocklist becomes a sandbox escape.

Attack Vector

An authenticated attacker submits a malicious flow or custom component through the Langflow web interface or REST API. The payload uses Python constructs not present in the blocklist to reach os.system, subprocess, or equivalent execution primitives. When the flow is validated and executed, the payload runs on the Langflow server.

The vulnerability manifests inside Langflow's custom-component validation path. See the IBM Support Page for vendor technical details and remediation guidance.

Stop Remote Code Execution Before It Runs

Singularity™ Endpoint detects and blocks exploit attempts at machine speed, including zero-days and novel attack techniques.

Stop Exploit Attempts

Detection Methods for CVE-2026-97655

Indicators of Compromise

  • Unexpected child processes spawned by the Langflow Python process, such as sh, bash, curl, or wget.
  • Outbound network connections from the Langflow host to unfamiliar destinations shortly after flow creation or edit events.
  • New or modified custom components containing obfuscated attribute lookups, __class__ chains, or dynamic __import__ calls.
  • Langflow API requests to component or flow endpoints from unusual source addresses or service accounts.

Detection Strategies

  • Monitor process trees under the Langflow service account and alert on execution of shells, interpreters, or network utilities.
  • Inspect stored flows and components for Python patterns that bypass denylists, including getattr, __subclasses__, and base64-decoded exec calls.
  • Correlate authentication events with component-upload and flow-run API calls to identify low-privilege accounts attempting code submission.

Monitoring Recommendations

  • Enable audit logging on all Langflow create, update, and run endpoints and forward logs to a centralized SIEM.
  • Baseline normal outbound traffic from Langflow hosts and alert on deviations.
  • Track file writes under Langflow working directories and container filesystems for unexpected binaries or scripts.

How to Mitigate CVE-2026-97655

Immediate Actions Required

  • Upgrade IBM Langflow OSS to a fixed version above 1.12.2 as published on the IBM Support Page.
  • Restrict network access to the Langflow API so that only trusted administrators can reach component and flow endpoints.
  • Rotate any API keys, model credentials, and secrets accessible to the Langflow process if exposure is suspected.
  • Review all stored custom components and flows for unauthorized or suspicious Python code.

Patch Information

IBM has published remediation guidance on the IBM Support Page. Administrators should apply the vendor-supplied fixed release for Langflow OSS. Versions 1.0.0 through 1.12.2 remain vulnerable until upgraded.

Workarounds

  • Disable custom component uploads and restrict flow authoring to a small set of trusted administrator accounts.
  • Run Langflow inside a hardened container with a read-only filesystem, no outbound internet access, and a non-root user.
  • Place Langflow behind an authenticating reverse proxy and enforce multi-factor authentication for all users.
  • Remove or scope down credentials and service tokens mounted into the Langflow runtime to the minimum necessary.
bash
# Example: restrict Langflow container with minimal privileges
docker run --rm \
  --read-only \
  --cap-drop=ALL \
  --security-opt=no-new-privileges \
  --user 10001:10001 \
  --network langflow-internal \
  -p 127.0.0.1:7860:7860 \
  langflowai/langflow:<fixed-version>

Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.

Sources: National Vulnerability Database, CISA KEV, Exploit-DB

Experience the Most Advanced Cybersecurity Platform

See how the world’s most intelligent, autonomous cybersecurity platform can protect your organization today and into the future.