Attack ChainAIDerived
Hypothetical attack flow derived from CVE metadata – not a confirmed exploit chain
-
AccessReach public Langflow endpoint
-
DeliveryIdentify public workflow with Python component
-
ExploitInvoke public execution path
-
ExecutionPre-existing Python runs in server process
-
ImpactCode executes as flow owner
Vulnerability Assessment
| Weakness |
CWE
OWASP Top 10:2025
A05:2025 - Injection
ASVS 5.0
V1 - Encoding and Sanitization Code injection prevention, sandbox exec
MITRE ATT&CK
2 techniques across 2 tactics: Execution, Initial Access The technique list with ATT&CK links and the kill-chain view is part of the Basic plan and up. |
| Exploitation Conditions | Exploitation requires a public workflow on the Langflow OSS server that already contains stored Python component code; the attacker does not inject new code but triggers this pre-existing component through a public execution path, which is supposed to be limited to server-trusted component code. The concrete prerequisite is therefore an exposed Langflow instance (versions 1.0.0 through 1.12.4) hosting at least one publicly accessible flow that embeds a Python component. The limiting factor is this deployment shape: instances with no public workflows, or no Python components in public flows, do not present the trigger. The CVSS vector indicates no authentication (PR:N) and no user interaction (UI:N) are required to invoke the public path. The code runs under the flow owner's runtime context within the server process, so impact scope is tied to that account's privileges. |
| Impact | Signals conflict and must be weighed carefully. The input CVSS of 9.8 (AV:N/AC:L/PR:N/UI:N, C:H/I:H/A:H) frames this as a critical remote, unauthenticated, low-complexity RCE, and the vuln.today Decision label is act_now with a patch already available from IBM - both arguing for urgent remediation. Against that, the CISA SSVC framework reports Exploitation: none (no observed exploitation) and Technical Impact: partial (not total), while flagging Automatable: yes. No EPSS score, no CISA KEV listing, and no PoC reference were supplied. The honest read: exploitation status is 'no public exploit identified at time of analysis,' so the act_now urgency rests on the combination of critical technical severity, the Automatable:yes signal, and an available fix - not on evidence of in-the-wild activity. The SSVC 'partial' technical impact is in tension with the CVSS C:H/I:H/A:H triad; resolve in favor of treating this as a genuine priority because the flaw bypasses an intended trust boundary to run Python under the flow owner's context, but note that real-world exposure depends on the precondition below rather than being universal. This is a real priority for anyone running an affected, publicly exposed Langflow instance, not merely a high-CVSS-but-low-real-risk issue - though the absence of exploitation and PoC data means it is not a confirmed emergency in the field. |
| Exploit ScenarioDerived | A remote attacker reaches a publicly exposed Langflow OSS instance (1.0.0-1.12.4) and invokes a public workflow that already contains a stored Python component. Because the public execution path fails to restrict execution to server-trusted component code, the attacker triggers that pre-existing Python to run inside the server process under the flow owner's runtime context, achieving code execution. No public exploit code or PoC reference was provided in the input. |
Get an e-mail when this CVE is exploited, patched or re-scored. Free account.
Alert meAnalysisAIv2 – Oct 30, 2026
Unauthenticated remote code execution in IBM Langflow OSS 1.0.0 through 1.12.4 lets a remote attacker reach a public execution path and trigger Python component code that is already stored inside a publicly accessible workflow, so the code runs inside the server process under the workflow owner's runtime context. The attacker does not inject new code; the flaw is a trust-boundary bypass in which the public path, intended to run only server-trusted component code, instead executes owner-supplied stored components. The deployment shape is the limiting factor - an exposed instance must host at least one public flow embedding a Python component - and no public exploit code has been identified at time of analysis, although the CVSS 3.1 vector (AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H, 9.8) shows no authentication or user interaction is needed to invoke the path, and the priority decision is act_now.
Earlier versions of this analysis (1)
IBM Langflow OSS 1.0.0 through 1.12.4 allows remote attackers to trigger pre-existing stored Python component code inside a public workflow, causing it to execute in the flow owner's runtime context within the server process and bypassing the trust boundary that normally restricts public execution paths to server-trusted component code. Per the CVSS vector (PR:N/UI:N), no authentication or interaction is needed once such a public workflow exists, though the flaw is only reachable in deployments that both expose a workflow publicly and have Python component code stored in it; impact is bounded to the flow owner's runtime context (C:L/I:L). It is a moderate-severity, code-injection class issue (CWE-94) to be handled as a routine prioritized patch this month; no public exploit code is identified in the available data and it is not in CISA KEV.
RemediationAI
A vendor patch is available per the IBM advisory at https://www.ibm.com/support/pages/node/7293016; the fixed version is not stated in the supplied data, so read the advisory for the exact release that closes CVE-2026-18894 and upgrade Langflow OSS 1.0.0 through 1.12.4 to it, treating this as act_now priority given the unauthenticated CVSS 9.8 vector. If upgrading cannot be done immediately, reduce exposure by removing public visibility from any workflow that contains a Python component (or removing the Python component from flows that must stay public) - the trade-off is that previously shared flows stop being reachable by external users - and, if public flows must remain, restrict network reachability to the Langflow server with allowlists or a reverse proxy so only intended clients can hit public execution paths, accepting that this weakens the sharing model the feature exists for. Additionally, run flows under owner accounts holding the least privilege feasible, since the triggered component executes with the flow owner's runtime context inside the server process, which limits blast radius but does not prevent execution. No vendor workaround beyond updating is documented in the available sources.
Affected VersionsENISA
Affected Packages
Source: OSV (Open Source Vulnerabilities)
Affects your stack
Affects 1 package in euvd
Checking this CVE against your own stacks and servers is part of the Basic plan and up.
Technical ContextAI · show details
Langflow is a low-code builder for LLM/AI agent flows in which a workflow is a graph of components, and Python components are stored code that executes server-side when the flow runs. CWE-94 (Improper Control of Generation of Code / code injection) here is not an input-to-eval injection in the classic sense: the malicious code is already stored in the workflow, and the defect is an access-control and trust-boundary failure that allows a public, unauthenticated execution path to reach that stored Python component instead of restricting public execution to server-trusted component code. Because the component runs within the Langflow server process under the flow owner's runtime context, the impact ceiling tracks the privileges and reachable resources of that owner account rather than the server process as a whole, which is why the technical impact is characterized as partial even though the CVSS impact sub-score is high across confidentiality, integrity and availability. The CPE entry cpe:2.3:a:ibm:langflow_oss:*:*:*:*:*:*:*:* carries no version constraints in the feed, so version scoping comes from the EUVD data (1.0.0 through 1.12.4) rather than from the CPE string. The exact request sequence and endpoint used to invoke the public path are not described in the available sources.
Know when Langflow Oss gets its next vulnerability
- Watch products and get an email when something new is published
- Check this CVE against your own stack and servers
- Exploit and PoC references, threat intelligence from 22 sources
No credit card · 7-day full trial
Severity by source
Sources disagree (Medium–Critical)AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H
Public execution path is reachable remotely without authentication or interaction (AV:N/AC:L/PR:N/UI:N); Python runs in-process under the flow owner yielding full host impact within that authority (S:U, C/I/A:H).
vuln.today treats the vendor’s rating as authoritative. A higher third-party CVSS (e.g. CISA-ADP) is shown for transparency but does not drive the headline severity.
CVSS VectorNVD
Lifecycle Timeline
3DescriptionNVD
IBM Langflow OSS 1.0.0 through 1.12.4 could allow a remote attacker to trigger pre-existing stored Python component code in a public workflow, executing it under the flow owner's runtime context within the server process. This bypasses the intended trust boundary where public execution paths are restricted to server-trusted component code.
Executive ViewAI
IBM Langflow OSS versions 1.0.0 through 1.12.4 contain a critical unauthenticated remote code execution vulnerability that allows attackers to execute arbitrary Python code on the server through public execution paths, affecting any organization with an internet-exposed instance hosting a public flow with a Python component. This flaw bypasses trust boundaries, enabling full server compromise without authentication or user interaction, posing severe risks to data confidentiality, integrity, and availability. Leadership should prioritize immediate patching to prevent potential exploitation.
Apply vendor patch.
Threat Intelligence Overview
How this decision was reached
The engine reads each signal below and applies a fixed list of rules from top to bottom. The first rule that matches sets the recommended action – like a statement, every line is shown.
What every action level means
SSVC AssessmentENISA
Priority Score
Attack Technique
Code injection occurs when an application accepts user input and passes it directly into a language interpreter or evaluator without proper sanitization.
Remote Code Execution represents the critical moment when an attacker successfully runs arbitrary code on a target system without physical access.
More in Langflow Oss
View allUnauthenticated remote code execution in IBM Langflow OSS versions 1.0.0 through 1.10.0 lets any network-reachable attac
Authenticated command injection in IBM Langflow OSS 1.0.0 through 1.11.1 lets any logged-in flow user escalate to arbitr
Code injection in IBM Langflow OSS versions 1.0.0 through 1.10.0 lets an authenticated user execute arbitrary operating-
Unauthenticated remote code execution in IBM Langflow OSS 1.0.0 through 1.12.2 lets any attacker who can reach the servi
Remote code execution in IBM Langflow OSS 1.0.0 through 1.11.1 allows a network-accessible authenticated attacker with l
IBM Langflow OSS 1.0.0 through 1.12.8 allows unauthenticated remote attackers to bypass security restrictions through im
Unauthenticated remote code execution in IBM Langflow OSS versions 1.0.0 through 1.9.3 allows attackers to fully comprom
Unauthenticated remote code execution in IBM Langflow OSS 1.0.0 through 1.10.1 allows remote attackers to run arbitrary
Authentication bypass in IBM Langflow OSS 1.0.0 through 1.10.0 lets an unauthenticated remote attacker retrieve a long-l
Remote code execution in IBM Langflow OSS versions 1.0.0 through 1.11.5 allows an unauthenticated remote attacker to run
Remote OS command execution in IBM Langflow OSS (versions 1.0.0 through 1.11.5) lets remote attackers run arbitrary oper
Path traversal in IBM Langflow OSS 1.0.0-1.10.3 exposes arbitrary server filesystem files via a crafted MCP `resources/r
EUVD-2026-113159
GHSA-pf68-vvvh-6f8f
NVD
CVE.org
EPSS
Vulnrichment
Exploit-DB