Langflow Smart Transform Code Injection Vulnerability Analysis

Vulnerability Overview Versions 1.3.0 to 1.10.2 of Langflow contain a code injection vulnerability in the Smart Transform (LambdaFilterComponent) component. Smart Transform incorporates flow author instructions and input data previews into prompts, requesting the Large Language Model (LLM) to generate a Python lambda function. It extracts a single line of lambda from the model's response, applies only syntax format validation, evaluates it using Python’s full built-in functions, and executes the resulting function within the Langflow process. Malicious flow authors can directly exploit this vulnerability via the Instructions field. In deployed environments, if exposed flows pass attacker-controlled content to Smart Transform, attackers may also exploit this indirectly through prompt injection, depending on the configured model. Affected Versions Affected versions: >=1.3.0, <1.10.3 Fixed version: 1.10.3 Remediation Primary fix: Addressed in PR #13530 and released in Langflow 1.11.0. Backport fix for 1.10.3: Addressed in PR #14073. Recommendation: Users should upgrade to Langflow 1.10.3 or later. Temporary Workarounds Before upgrading, the following measures can be taken: Remove Smart Transform from runnable flows. Restrict flow creation, editing, and execution to trusted users. Do not route untrusted or externally controlled data through Smart Transform. Restrict the file system, network, and credential access of the Langflow process. POC Code For example, an attacker might attempt to make the model return: This expression satisfies the vulnerability's format checks. The call creates a lambda with Python’s default built-in functions, and the subsequent call (within ) executes the command. Successful exploitation allows code execution with the privileges of the Langflow service process. This may expose or modify credentials, files, application data, and network resources accessible to that process, and potentially impact other tenants in a shared deployment.

Goal Reached Thanks to every supporter — we hit 100%!

Langflow Smart Transform Code Injection Vulnerability Analysis

More from this source