AI Threat AlertCVE-2026-19303: Langflow: path traversal enables arbitrary file deletion
HIGHThis vulnerability lets a low-privileged authenticated user of an IBM Langflow OSS deployment (versions 1.0.0 through 1.11.2) delete arbitrary files or directories on the host, due to a classic path traversal flaw (CWE-22) in file-handling logic. For a CISO, the real risk lies in Langflow's growing footprint as the visual builder behind LLM agent and RAG pipelines: a single malicious or compromised low-privilege account, plausible in any shared or multi-tenant Langflow deployment, could wipe flow definitions, configuration files, or logs, silently disrupting production AI workflows or covering an attacker's tracks. There is no evidence of active exploitation, no CISA KEV listing, no published EPSS score, and no public exploit code or Nuclei template, so this is not an imminent mass-exploitation risk, but the low attack complexity (AC:L) and lack of required user interaction mean it is trivial to weaponize once someone has any authenticated foothold. Teams running Langflow should upgrade past 1.11.2 as soon as IBM publishes a fixed build, audit which accounts hold write/delete access to Langflow's file operations, and add monitoring for unexpected file deletions in Langflow's working directories in the interim.
What is the risk?
CVSS 8.1 (High) reflects a network-exploitable, low-complexity flaw requiring only low privileges and no user interaction, with full integrity and availability impact (C:N/I:H/A:H) — the attacker cannot read data but can destroy it. The requirement for authentication (PR:L) meaningfully caps exploitability versus a fully unauthenticated bug, and the absence of KEV listing, EPSS score, public PoC, or Nuclei coverage indicates no observed exploitation in the wild today. Risk should be weighted higher for organizations running multi-tenant or shared Langflow instances where low-privilege accounts (e.g., junior developers, contractors, or self-service users) are common, since that is exactly the access level needed to trigger the flaw.
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 to the IBM-patched Langflow release once available (track the linked IBM support advisory for the fixed version number, as none is specified beyond 'through 1.11.2'). Until patched, restrict Langflow instances to trusted, need-to-know users only and avoid exposing low-privilege or self-service accounts on shared deployments. Apply least-privilege at the OS/filesystem level (run Langflow under a service account with write access scoped strictly to its own data directory, using containerization or chroot/jail where possible) so that even a successful traversal cannot reach files outside that boundary. Add detection for anomalous file-deletion patterns or unexpected 404/500 errors on file-management endpoints, and review Langflow access logs for
../or encoded traversal sequences in requests.
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-2026-19303?
This vulnerability lets a low-privileged authenticated user of an IBM Langflow OSS deployment (versions 1.0.0 through 1.11.2) delete arbitrary files or directories on the host, due to a classic path traversal flaw (CWE-22) in file-handling logic. For a CISO, the real risk lies in Langflow's growing footprint as the visual builder behind LLM agent and RAG pipelines: a single malicious or compromised low-privilege account, plausible in any shared or multi-tenant Langflow deployment, could wipe flow definitions, configuration files, or logs, silently disrupting production AI workflows or covering an attacker's tracks. There is no evidence of active exploitation, no CISA KEV listing, no published EPSS score, and no public exploit code or Nuclei template, so this is not an imminent mass-exploitation risk, but the low attack complexity (AC:L) and lack of required user interaction mean it is trivial to weaponize once someone has any authenticated foothold. Teams running Langflow should upgrade past 1.11.2 as soon as IBM publishes a fixed build, audit which accounts hold write/delete access to Langflow's file operations, and add monitoring for unexpected file deletions in Langflow's working directories in the interim.
Is CVE-2026-19303 actively exploited?
No confirmed active exploitation of CVE-2026-19303 has been reported, but organizations should still patch proactively.
How to fix CVE-2026-19303?
Upgrade to the IBM-patched Langflow release once available (track the linked IBM support advisory for the fixed version number, as none is specified beyond 'through 1.11.2'). Until patched, restrict Langflow instances to trusted, need-to-know users only and avoid exposing low-privilege or self-service accounts on shared deployments. Apply least-privilege at the OS/filesystem level (run Langflow under a service account with write access scoped strictly to its own data directory, using containerization or chroot/jail where possible) so that even a successful traversal cannot reach files outside that boundary. Add detection for anomalous file-deletion patterns or unexpected 404/500 errors on file-management endpoints, and review Langflow access logs for `../` or encoded traversal sequences in requests.
What systems are affected by CVE-2026-19303?
This vulnerability affects the following AI/ML architecture patterns: agent frameworks, LLM orchestration pipelines, RAG pipelines.
What is the CVSS score for CVE-2026-19303?
CVE-2026-19303 has a CVSS v3.1 base score of 8.1 (HIGH). The EPSS exploitation probability is 0.61%.
What is the AI security impact?
Affected AI Architectures
MITRE ATLAS Techniques
AML.T0029Denial of AI ServiceCompliance 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 delete arbitrary local files or directories due to improper limitation of a pathname to a restricted directory.
Exploitation Scenario
An attacker who has obtained (or been granted) a low-privilege authenticated account on a shared Langflow deployment — for example, a contractor account or a self-registered user in an internal AI platform — sends a crafted request to one of Langflow's file-management or flow-deletion API endpoints, substituting a path traversal payload (e.g., sequences like `../../`) for the expected filename or flow ID. Because the endpoint fails to properly restrict the resolved path to its intended directory, the deletion request executes against an arbitrary file or directory elsewhere on the host, letting the attacker destroy other tenants' flows, wipe configuration files, or remove logs to cover further malicious activity — all without needing elevated privileges or any user interaction from a victim.
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:N/I:H/A:HReferences
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