CVE-2026-19301: Langflow SSRF — MEDIUM

CISO Take

IBM Langflow OSS versions 1.0.0 through 1.11.2 contain a server-side request forgery flaw (CWE-918) that lets any authenticated user with only low privileges coerce the Langflow server into making outbound HTTP requests on their behalf. Because Langflow is a visual builder for LLM and agent pipelines, this isn't a fringe admin panel — it's the orchestration layer itself, and any multi-tenant or team deployment where non-admin users can create or edit flows is exposed. The CVSS 5.0 (medium) score reflects a confidentiality-only impact with no integrity or availability loss, and there's no EPSS score, no CISA KEV listing, and no public exploit or Nuclei template yet, so this looks like disclosure-driven rather than actively-exploited. Even so, SSRF in a server that typically runs inside cloud VPCs is a classic pivot to cloud metadata endpoints (e.g., IMDS) and internal-only services, so treat it as a credential-theft precursor, not just an info leak. Action: upgrade Langflow past 1.11.2 per the IBM advisory, and in the meantime restrict which roles can create/edit flows with URL-accepting components, enforce IMDSv2 and network egress controls on the host running Langflow, and monitor outbound traffic from the Langflow process for requests to RFC1918 ranges or 169.254.169.254.

Sources:NVDibm.comATLAS

What is the risk?

Network-exploitable, low attack complexity, and requires no user interaction, but does require an authenticated (low-privilege) account — this caps exploitability to insiders, compromised accounts, or deployments with open/self-service registration. Impact is scoped to confidentiality-low with a scope change (S:C), consistent with classic SSRF: the attacker can't directly read arbitrary files or execute code, but can induce the server to make requests to internal-only endpoints and potentially learn about internal topology or leak partial response data. No EPSS score, no CISA KEV entry, and no public PoC/Nuclei template currently exist, so real-world exploitation activity appears low today. Risk rises sharply for any Langflow deployment that (a) allows self-signup or has many low-trust users, (b) runs inside a cloud VPC with an accessible metadata service, or (c) exposes internal services reachable from the Langflow host.

How does the attack unfold?

Initial Access

Attacker obtains or already holds a low-privilege authenticated Langflow account (e.g., standard team member or self-registered user).

AML.T0012

Exploitation

Attacker creates or edits a flow, configuring a URL-accepting component (HTTP request/webhook node) to target an internal or cloud-metadata endpoint instead of a legitimate external API.

AML.T0049

Reconnaissance

The flow executes server-side, causing the Langflow backend to issue the forged request with its own network privileges, reaching internal services normally inaccessible to the attacker.

AML.T0075

Impact

Response data from the internal request (e.g., service banners, partial credentials, internal topology) is surfaced back to the attacker via the flow's output, informing further compromise.

AML.T0025

Initial Access

Attacker obtains or already holds a low-privilege authenticated Langflow account (e.g., standard team member or self-registered user).

AML.T0012

Exploitation

Attacker creates or edits a flow, configuring a URL-accepting component (HTTP request/webhook node) to target an internal or cloud-metadata endpoint instead of a legitimate external API.

AML.T0049

Reconnaissance

The flow executes server-side, causing the Langflow backend to issue the forged request with its own network privileges, reaching internal services normally inaccessible to the attacker.

AML.T0075

Impact

Response data from the internal request (e.g., service banners, partial credentials, internal topology) is surfaced back to the attacker via the flow's output, informing further compromise.

AML.T0025

What systems are affected?

Package	Ecosystem	Vulnerable Range	Patched
Langflow	pip	—	No patch
156.2KPushed yesterday18% patched~66d to patchFull package profile →

Do you use Langflow? You're affected.

How severe is it?

CVSS 3.1

6.5 / 10

EPSS

0.4%

chance of exploitation in 30 days

Higher than 34% of all CVEs

Source: EPSS v3 — FIRST.org

Exploitation Status

No known exploitation

Sophistication

Moderate

What is the attack surface?

AVNetwork

ACLow

PRLow

UINone

SUnchanged

CHigh

INone

ANone

What should I do?

1 step

1) Upgrade Langflow OSS beyond 1.11.2 per the IBM advisory (https://www.ibm.com/support/pages/node/7285639) as soon as a fixed release is confirmed. 2) Until patched, restrict flow-creation/edit permissions to trusted users only — treat 'can create/edit flows' as an effectively high-privilege capability given this flaw. 3) Enforce network egress controls from the host/container running Langflow: block outbound access to RFC1918/link-local ranges and the cloud metadata IP (169.254.169.254) unless explicitly required, and require IMDSv2 on AWS to blunt metadata-service SSRF. 4) Segment Langflow from sensitive internal services (vector DBs, internal APIs, secrets managers) via network policy rather than relying on application-layer trust. 5) Detection: monitor Langflow server outbound connections for requests to internal/private IP ranges or metadata endpoints that don't match expected external API integrations; alert on unusual response sizes/timing from flow-triggered HTTP components.

What does CISA's SSVC say?

DecisionTrack

Exploitationnone

AutomatableNo

Technical Impactpartial

Source: CISA Vulnrichment (SSVC v2.0). Decision based on the CISA Coordinator decision tree.

How is it classified?

Data LeakageFrameworkAgentAML.T0049- Exploit Public-Facing Application AML.T0075- Cloud Service Discovery

Which compliance frameworks are affected?

This CVE is relevant to:

EU AI Act

Article 15-Accuracy, robustness and cybersecurity

ISO 42001

A.6.2.6-AI system security controls

NIST AI RMF

MEASURE 2.7-AI system security and resilience are evaluated and documented

OWASP LLM Top 10

LLM07-Insecure Plugin Design

Frequently Asked Questions

What is CVE-2026-19301?

IBM Langflow OSS versions 1.0.0 through 1.11.2 contain a server-side request forgery flaw (CWE-918) that lets any authenticated user with only low privileges coerce the Langflow server into making outbound HTTP requests on their behalf. Because Langflow is a visual builder for LLM and agent pipelines, this isn't a fringe admin panel — it's the orchestration layer itself, and any multi-tenant or team deployment where non-admin users can create or edit flows is exposed. The CVSS 5.0 (medium) score reflects a confidentiality-only impact with no integrity or availability loss, and there's no EPSS score, no CISA KEV listing, and no public exploit or Nuclei template yet, so this looks like disclosure-driven rather than actively-exploited. Even so, SSRF in a server that typically runs inside cloud VPCs is a classic pivot to cloud metadata endpoints (e.g., IMDS) and internal-only services, so treat it as a credential-theft precursor, not just an info leak. Action: upgrade Langflow past 1.11.2 per the IBM advisory, and in the meantime restrict which roles can create/edit flows with URL-accepting components, enforce IMDSv2 and network egress controls on the host running Langflow, and monitor outbound traffic from the Langflow process for requests to RFC1918 ranges or 169.254.169.254.

Is CVE-2026-19301 actively exploited?

No confirmed active exploitation of CVE-2026-19301 has been reported, but organizations should still patch proactively.

How to fix CVE-2026-19301?

1) Upgrade Langflow OSS beyond 1.11.2 per the IBM advisory (https://www.ibm.com/support/pages/node/7285639) as soon as a fixed release is confirmed. 2) Until patched, restrict flow-creation/edit permissions to trusted users only — treat 'can create/edit flows' as an effectively high-privilege capability given this flaw. 3) Enforce network egress controls from the host/container running Langflow: block outbound access to RFC1918/link-local ranges and the cloud metadata IP (169.254.169.254) unless explicitly required, and require IMDSv2 on AWS to blunt metadata-service SSRF. 4) Segment Langflow from sensitive internal services (vector DBs, internal APIs, secrets managers) via network policy rather than relying on application-layer trust. 5) Detection: monitor Langflow server outbound connections for requests to internal/private IP ranges or metadata endpoints that don't match expected external API integrations; alert on unusual response sizes/timing from flow-triggered HTTP components.

What systems are affected by CVE-2026-19301?

This vulnerability affects the following AI/ML architecture patterns: agent frameworks, LLM orchestration/workflow pipelines, webhook and API integration components.

What is the CVSS score for CVE-2026-19301?

CVE-2026-19301 has a CVSS v3.1 base score of 6.5 (MEDIUM). The EPSS exploitation probability is 0.39%.

What is the AI security impact?

Affected AI Architectures

agent frameworksLLM orchestration/workflow pipelineswebhook and API integration components

MITRE ATLAS Techniques

AML.T0049Exploit Public-Facing Application

AML.T0075Cloud Service Discovery

Compliance Controls Affected

EU AI Act: Article 15

ISO 42001: A.6.2.6

NIST AI RMF: MEASURE 2.7

OWASP LLM Top 10: LLM07

What are the technical details?

Original Advisory

IBM Langflow OSS 1.0.0 through 1.11.2 could allow a remote authenticated attacker to obtain sensitive information due to server-side request forgery.

Exploitation Scenario

An attacker with a low-privilege Langflow account (e.g., a standard team member in a multi-tenant deployment, or an account obtained via open self-registration) creates or edits a flow and adds a component that accepts a URL parameter — such as an HTTP request node or custom API/webhook integration. Instead of a legitimate external endpoint, the attacker points it at the cloud instance metadata service (169.254.169.254) or an internal-only admin API reachable from the Langflow host. When the flow executes, the Langflow backend performs the request server-side with its own network privileges, and the (partial) response is surfaced back to the attacker through the flow's output or logs — potentially revealing cloud IAM credentials, internal service banners, or other data that should never be reachable from a low-trust user account.

Weaknesses (CWE)

CWE-918Server-Side Request ForgeryPrimary CWE-918Server-Side Request Forgery

CWE-918 — Server-Side Request Forgery: The web server receives a URL or similar request from an upstream component and retrieves the contents of this URL, but it does not sufficiently ensure that the request is being sent to the expected destination.

[Implementation, Operation] If there is a known list of applications, IP, hostname, and/or domains to which the product is allowed to communicate, then use allowlists to ensure that outgoing requests are restricted to these known entities. The allowlists could be in the application code or possibly enforced by firewalls or network segmentation [REF-1511].
[Implementation, Operation] If the product is expected to be able to connect to arbitrary IP addresses or domains, consider using a denylist that includes private network addresses, internal domain names, etc. The denylist could be in the application code or possibly enforced by firewalls or network segmentation [REF-1511].

Source: MITRE CWE corpus.