CVE-2026-104312: Langflow: ZIP path traversal enables file write on victim

MEDIUM

Published October 29, 2026

CISO Take

IBM Langflow OSS 1.0.0 through 1.12.3 can be made to generate ZIP archives whose entry paths escape the extraction directory (a classic Zip Slip, CWE-22). An authenticated attacker can produce such an archive, and if a victim extracts it with a non-sanitizing tool, arbitrary files can be written on the victim's host. The risk is limited: the CVSS score is 3.5 (low), the attack requires a low-privileged account plus a victim who must manually extract the archive, there is no public exploit, no Nuclei template, and it is not in CISA KEV. EPSS data is not yet available, so treat exploitation likelihood as low but unconfirmed. Upgrade to a fixed Langflow release per the IBM advisory (support node 7293015), and instruct staff to extract exported Langflow archives only with tools that reject paths outside the target directory.

Sources:NVDibm.comATLAS

What is the risk?

Low overall risk. The vector is network-based with low complexity, but it needs an authenticated account and user interaction, and the impact is limited to integrity (I:L) with no confidentiality or availability effect. The real exposure sits on the victim's workstation or build host rather than the Langflow server, and only materializes if a non-sanitizing extractor is used. Risk rises in shared, multi-tenant Langflow deployments where untrusted users hold accounts and where administrators routinely download and unpack exported flows or project bundles.

How does the attack unfold?

Initial Access

Attacker authenticates to a Langflow instance with a low-privileged account.

AML.T0012

Payload Crafting

Attacker triggers server-side generation of a ZIP archive whose entry paths traverse outside the extraction directory.

User Execution

Victim downloads the archive and extracts it with a non-sanitizing tool.

AML.T0011

Impact

Arbitrary files are written on the victim's host, tampering with local files and potentially enabling further compromise.

Initial Access

Attacker authenticates to a Langflow instance with a low-privileged account.

AML.T0012

Payload Crafting

Attacker triggers server-side generation of a ZIP archive whose entry paths traverse outside the extraction directory.

User Execution

Victim downloads the archive and extracts it with a non-sanitizing tool.

AML.T0011

Impact

Arbitrary files are written on the victim's host, tampering with local files and potentially enabling further compromise.

What systems are affected?

Package	Ecosystem	Vulnerable Range	Patched
Langflow	pip	—	No patch
156.0KPushed 6d ago18% patched~66d to patchFull package profile →

Do you use Langflow? You're affected.

How severe is it?

CVSS 3.1

6.8 / 10

EPSS

0.2%

chance of exploitation in 30 days

Higher than 13% of all CVEs

Source: EPSS v3 — FIRST.org

Exploitation Status

No known exploitation

Sophistication

Trivial

What is the attack surface?

AVNetwork

ACLow

PRLow

UIRequired

SChanged

CNone

IHigh

ANone

What should I do?

1 step

1) Upgrade Langflow OSS to the fixed version identified in IBM's advisory (https://www.ibm.com/support/pages/node/7293015). 2) Until patched, restrict Langflow account creation and review which users can generate or export archives. 3) Extract downloaded archives only with tools that sanitize paths (current unzip, Python zipfile with path validation, or a sandboxed/containerized extraction) and never as root. 4) Inspect archives before extraction, for example with unzip -l, and reject entries containing ../ or absolute paths. 5) Monitor for unexpected file modifications outside designated extraction directories and review Langflow audit logs for unusual export activity.

How is it classified?

Supply ChainFrameworkAML.T0010.001- AI Software AML.T0011- User Execution

Which compliance frameworks are affected?

This CVE is relevant to:

ISO 42001

A.6.2.6-AI system operation and monitoring

NIST AI RMF

MANAGE 2.3-Response to previously unknown risks and vulnerability management

OWASP LLM Top 10

LLM03-Supply Chain

Frequently Asked Questions

What is CVE-2026-104312?

Is CVE-2026-104312 actively exploited?

No confirmed active exploitation of CVE-2026-104312 has been reported, but organizations should still patch proactively.

How to fix CVE-2026-104312?

1) Upgrade Langflow OSS to the fixed version identified in IBM's advisory (https://www.ibm.com/support/pages/node/7293015). 2) Until patched, restrict Langflow account creation and review which users can generate or export archives. 3) Extract downloaded archives only with tools that sanitize paths (current unzip, Python zipfile with path validation, or a sandboxed/containerized extraction) and never as root. 4) Inspect archives before extraction, for example with unzip -l, and reject entries containing ../ or absolute paths. 5) Monitor for unexpected file modifications outside designated extraction directories and review Langflow audit logs for unusual export activity.

What systems are affected by CVE-2026-104312?

This vulnerability affects the following AI/ML architecture patterns: agent frameworks, LLM application development platforms, RAG pipelines, flow export and sharing workflows.

What is the CVSS score for CVE-2026-104312?

CVE-2026-104312 has a CVSS v3.1 base score of 6.8 (MEDIUM). The EPSS exploitation probability is 0.22%.

What is the AI security impact?

Affected AI Architectures

agent frameworksLLM application development platformsRAG pipelinesflow export and sharing workflows

MITRE ATLAS Techniques

AML.T0010.001AI Software

AML.T0011User Execution

Compliance Controls Affected

ISO 42001: A.6.2.6

NIST AI RMF: MANAGE 2.3

OWASP LLM Top 10: LLM03

What are the technical details?

Original Advisory

IBM Langflow OSS 1.0.0 through 1.12.3 could allow an authenticated attacker to produce a server-generated ZIP archive containing entry paths that traverse outside the intended extraction directory. Extraction of the archive by a victim using a non-sanitizing tool could result in arbitrary file write on the victim's host.

Exploitation Scenario

An attacker with a low-privileged account on a shared Langflow instance crafts a flow or project so the server-generated export ZIP contains entries such as ../../.bashrc or a path into a startup directory. The attacker persuades an administrator or colleague to download and unpack the shared export, for example by sending a link to a useful template flow. If the victim uses a naive extractor, the malicious entries are written outside the intended folder, tampering with files on the victim's host and potentially setting up later code execution.

Weaknesses (CWE)

CWE-22Improper Limitation of a Pathname to a Restricted Directory ('Path Traversal')Primary CWE-22Improper Limitation of a Pathname to a Restricted Directory ('Path Traversal')

CWE-22 — Improper Limitation of a Pathname to a Restricted Directory ('Path Traversal'): The product uses external input to construct a pathname that is intended to identify a file or directory that is located underneath a restricted parent directory, but the product does not properly neutralize special elements within the pathname that can cause the pathname to resolve to a location that is outside of the restricted directory.

[Implementation] Assume all input is malicious. Use an "accept known good" input validation strategy, i.e., use a list of acceptable inputs that strictly conform to specifications. Reject any input that does not strictly conform to specifications, or transform it into something that does. When performing input validation, consider all potentially relevant properties, including length, type of input, the full range of acceptable values, missing or extra inputs, syntax, consistency across related fields, and conformance to business rules. As an example of business rule logic, "boat" may be syntactically valid because it only contains alphanumeric characters, but it is not valid if the input is only expected to contain colors such as "red" or "blue." Do not rely exclusively on looking for malicious or malformed inputs. This is likely to miss at least one undesirable input, especially if the code's environment changes. This can give attackers enough room to bypass the intended validation. However, denylis
[Architecture and Design] For any security checks that are performed on the client side, ensure that these checks are duplicated on the server side, in order to avoid CWE-602. Attackers can bypass the client-side checks by modifying values after the checks have been performed, or by changing the client to remove the client-side checks entirely. Then, these modified values would be submitted to the server.

Source: MITRE CWE corpus.