Langflow

IBM Langflow OSS versions 1.0.0 through 1.12.2 contain an improper access control vulnerability (CVE-2026-104335) that allows an authenticated remote attacker to execute arbitrary code.

Langflow versions 1.10.0 and earlier use a non-cryptographic PRNG to derive Fernet keys from short SECRET_KEYs, allowing unauthenticated attackers who obtain the secret key file to perform offline decryption of all stored user credentials.

Langflow contains multiple vulnerabilities that enable remote attackers to achieve remote code execution with administrative privileges and bypass existing security controls.

IBM Langflow OSS versions 1.0.0 through 1.10.1 are susceptible to unauthenticated remote code execution due to improper sanitization of environment variables in the MCP stdio launcher.

IBM Langflow OSS versions 1.0.0 through 1.10.0 contain a critical deserialization vulnerability (CVE-2026-8476) in its disk-based caching mechanism, which uses Python's unsafe `pickle.loads()` function without proper validation, allowing attackers to process malicious pickle payloads and achieve arbitrary code execution with the privileges of the Langflow server process, leading to complete system compromise.

An authenticated attacker can exploit CVE-2026-9135, a code injection vulnerability in IBM Langflow OSS versions 1.0.0 through 1.9.2, to bypass security controls and achieve arbitrary Python code execution on the backend through unvalidated dynamic CodeInput fields in the ToolGuard integration, potentially escalating privileges via cross-tenant flow manipulation.

A remote attacker can gain full administrative access to IBM Langflow OSS versions 1.0.0 through 1.10.0 by exploiting an improper authentication vulnerability. The /api/v1/login/auto_login endpoint, when the default AUTO_LOGIN configuration is enabled, issues long-lived superuser bearer tokens without requiring authentication. This allows an unauthenticated network attacker to obtain these tokens and achieve superuser privileges. Additionally, permissive Cross-Origin Resource Sharing (CORS) settings could expose these tokens to unintended origins, exacerbating the risk.

An unauthenticated file upload vulnerability, CVE-2026-55450, in Langflow versions prior to 1.9.1 allows remote attackers to upload arbitrary amounts of data to the server via the `/api/v1/upload/{flow_id}` endpoint, leading to disk space exhaustion and Denial-of-Service, additionally leaking the absolute file path in the server's response.

A critical unauthenticated Remote Code Execution (RCE) vulnerability (CVE-2026-48519) exists in Langflow versions up to and including 1.9.1, allowing attackers to execute arbitrary Python code by sending a crafted JSON payload to the `/api/v1/build_public_tmp` API endpoint of a publicly shared flow.

Langflow 1.3.0 contains a remote code execution vulnerability (CVE-2026-0770) due to untrusted input in the exec_globals parameter at the validate endpoint, allowing remote attackers to execute arbitrary code as root without authentication, as demonstrated by a public exploit.

An authenticated remote attacker can exploit a vulnerability in Langflow to perform a denial-of-service attack, impacting system availability.

A path traversal vulnerability exists in the Langflow Knowledge Bases API (`DELETE /api/v1/knowledge_bases`) that allows an authenticated attacker to delete arbitrary directories on the server's filesystem, leading to data loss and potential service disruption.

An authenticated remote attacker can exploit multiple unspecified vulnerabilities in Langflow to achieve arbitrary code execution.

Multiple vulnerabilities in Langflow allow an attacker to manipulate files, disclose sensitive information, or conduct cross-site scripting attacks.

An unrestricted file upload vulnerability in langflow-ai langflow versions up to 1.1.0 allows remote attackers to execute arbitrary code via the create_upload_file function in the API Endpoint.

IBM Langflow Desktop versions 1.6.0 through 1.8.2 is vulnerable to arbitrary code execution due to insecure deserialization of untrusted data, allowing an authenticated user to execute code on the system.

An anonymous or authenticated remote attacker can exploit multiple vulnerabilities in Langflow to disclose information, manipulate data, and execute cross-site scripting attacks.

An authenticated, remote attacker can exploit a vulnerability in Langflow to manipulate files, potentially leading to unauthorized data modification or application compromise.

Langflow versions 1.5.0 and earlier contain an IDOR vulnerability (CVE-2026-34046) that allows authenticated users to read, modify, and delete flows belonging to other users due to a missing ownership check, potentially exposing sensitive information and enabling unauthorized control over AI agent logic.

A vulnerability in Langflow allows an attacker to execute arbitrary code, potentially leading to system compromise.

Multiple vulnerabilities in Langflow could be exploited by an attacker to execute arbitrary program code, disclose information, and potentially manipulate data, leading to potential system compromise.

A critical remote code execution vulnerability, CVE-2026-33017, exists in Langflow AI pipelines prior to version 1.9.0 that allows an unauthenticated remote attacker to execute code with full server process privileges, impacting availability, integrity, and confidentiality.

Langflow versions 1.0.0 through 1.8.1 are vulnerable to an unauthenticated image retrieval vulnerability (CVE-2026-33484) that allows attackers to download any user's uploaded images without credentials in multi-tenant deployments by accessing the `/api/v1/files/images/{flow_id}/{file_name}` endpoint.