AI Threat AlertCVE-2026-19302: Langflow: symlink flaw leaks sensitive files
MEDIUMIBM Langflow OSS versions 1.0.0 through 1.11.2 fail to properly validate symbolic links, letting an authenticated attacker with only low privileges read files outside the directories the application intends to expose. Langflow is a widely used visual builder for LLM and agent pipelines, so this is not a fringe tool — a successful read could surface API keys, provider credentials, flow configuration secrets, or other tenants' project data stored on the host. There's no CISA KEV listing, no public exploit code, and no Nuclei template as of this writing, and EPSS data isn't yet available, so this reads as a low-noise but real risk rather than an emergency. Because exploitation requires network access plus a valid low-privilege account (CVSS 6.5, AC:L, PR:L, UI:N) and yields confidentiality impact only, prioritize it above routine patching but below anything actively exploited. Upgrade past 1.11.2 as soon as IBM/Langflow publishes a fixed release, and in the meantime audit which users hold accounts on multi-tenant Langflow instances and check for anomalous file-read patterns in application logs.
What is the risk?
Medium severity (CVSS 6.5) driven entirely by a high confidentiality impact with no integrity or availability effect. The attack requires network access, low privileges, and no user interaction, and complexity is low — so once an attacker has any account on a shared or multi-tenant Langflow deployment, exploitation is straightforward. Mitigating factors: no evidence of active exploitation, absence from CISA KEV, no public PoC or scanner template, and EPSS scoring unavailable, suggesting this is not yet being targeted in the wild. The realistic exposure is highest for organizations running Langflow as a shared/multi-user service (e.g., internal AI platform teams) rather than single-user local deployments.
How does the attack unfold?
What systems are affected?
| Package | Ecosystem | Vulnerable Range | Patched |
|---|---|---|---|
| Langflow | pip | — | No patch |
Do you use Langflow? You're affected.
How severe is it?
What is the attack surface?
What should I do?
1 stepUpgrade Langflow to the first version after 1.11.2 that addresses this CWE-22 symlink validation issue once IBM/Langflow publishes it (track https://www.ibm.com/support/pages/node/7285641 for the fix version). Until patched: restrict who can create low-privilege accounts on shared Langflow instances, run the Langflow service under a minimally privileged OS account with restricted filesystem access (chroot/containerize with a tightly scoped bind mount), avoid storing secrets in files readable by the service account (use a secrets manager or environment injection at the orchestrator level instead of
.env), and monitor application/file-access logs for path-traversal or symlink-creation patterns tied to file upload or flow-import features. Rotate any credentials that may have been exposed if you suspect exploitation.
What does CISA's SSVC say?
Source: CISA Vulnrichment (SSVC v2.0). Decision based on the CISA Coordinator decision tree.
How is it classified?
Which compliance frameworks are affected?
This CVE is relevant to:
Frequently Asked Questions
What is CVE-2026-19302?
IBM Langflow OSS versions 1.0.0 through 1.11.2 fail to properly validate symbolic links, letting an authenticated attacker with only low privileges read files outside the directories the application intends to expose. Langflow is a widely used visual builder for LLM and agent pipelines, so this is not a fringe tool — a successful read could surface API keys, provider credentials, flow configuration secrets, or other tenants' project data stored on the host. There's no CISA KEV listing, no public exploit code, and no Nuclei template as of this writing, and EPSS data isn't yet available, so this reads as a low-noise but real risk rather than an emergency. Because exploitation requires network access plus a valid low-privilege account (CVSS 6.5, AC:L, PR:L, UI:N) and yields confidentiality impact only, prioritize it above routine patching but below anything actively exploited. Upgrade past 1.11.2 as soon as IBM/Langflow publishes a fixed release, and in the meantime audit which users hold accounts on multi-tenant Langflow instances and check for anomalous file-read patterns in application logs.
Is CVE-2026-19302 actively exploited?
No confirmed active exploitation of CVE-2026-19302 has been reported, but organizations should still patch proactively.
How to fix CVE-2026-19302?
Upgrade Langflow to the first version after 1.11.2 that addresses this CWE-22 symlink validation issue once IBM/Langflow publishes it (track https://www.ibm.com/support/pages/node/7285641 for the fix version). Until patched: restrict who can create low-privilege accounts on shared Langflow instances, run the Langflow service under a minimally privileged OS account with restricted filesystem access (chroot/containerize with a tightly scoped bind mount), avoid storing secrets in files readable by the service account (use a secrets manager or environment injection at the orchestrator level instead of `.env`), and monitor application/file-access logs for path-traversal or symlink-creation patterns tied to file upload or flow-import features. Rotate any credentials that may have been exposed if you suspect exploitation.
What systems are affected by CVE-2026-19302?
This vulnerability affects the following AI/ML architecture patterns: agent frameworks, RAG pipelines, multi-tenant AI platform deployments.
What is the CVSS score for CVE-2026-19302?
CVE-2026-19302 has a CVSS v3.1 base score of 6.5 (MEDIUM). The EPSS exploitation probability is 0.66%.
What is the AI security impact?
Affected AI Architectures
MITRE ATLAS Techniques
AML.T0037Data from Local SystemCompliance Controls Affected
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 improper validation of symbolic links.
Exploitation Scenario
An attacker who has obtained (or was granted) a low-privilege account on a shared Langflow instance uploads or creates a flow artifact that is actually a symbolic link pointing to a sensitive path outside the intended project directory — for example targeting the server's `.env` file, another tenant's saved flow, or SSH/config files. When Langflow later processes, exports, or serves that artifact (e.g., via a file-download or flow-import endpoint) without resolving and validating the link target, it follows the symlink and returns the contents of the sensitive file to the attacker, who now holds LLM provider API keys or other tenants' pipeline configurations without ever needing elevated privileges.
Weaknesses (CWE)
CWE-22Improper Limitation of a Pathname to a Restricted DirectoryPrimaryCWE-22Improper Limitation of a Pathname to a Restricted DirectoryCWE-22 — Improper Limitation of a Pathname to a Restricted Directory: 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.
CVSS Vector
CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:N/A:NReferences
Timeline
Related Vulnerabilities
CVE-2026-1056110.0Langflow: auth bypass + unauthenticated RCE (CVSS 10)
Same package: langflowCVE-2026-1013410.0Langflow: unauthenticated RCE via tool_code injection
Same package: langflowCVE-2026-333099.9langflow: Path Traversal enables file access
Same package: langflowCVE-2026-552559.9Langflow: IDOR allows cross-user flow execution
Same package: langflowCVE-2026-78739.9Langflow: authenticated RCE enables credential theft
Same package: langflow