Tandoor Recipes is vulnerable to a stored cross-site scripting issue due to improper sanitization of user input in recipe step instructions. The bleach.clean() sanitizer whitelist includes the <style> tag, allowing authenticated users to inject arbitrary CSS that is persisted and served by the backend API. When clients render the instructions_markdown field as HTML without additional sanitization, the injected CSS executes in the client context. This can lead to UI manipulation, phishing, visual defacement, and data exfiltration via CSS techniques. The issue affects versions prior to 2.6.4 and is resolved in 2.6.4.

The vulnerability allows authenticated users to inject malicious CSS into recipe instructions, which is then served to other users or clients consuming the API. This can result in UI redressing attacks, phishing overlays, visual defacement, and CSS-based data exfiltration. The impact is limited to authenticated users injecting payloads and clients that render the affected content without additional sanitization. There is no indication of remote unauthenticated exploitation or direct code execution. The CVSS score of 5.4 reflects a medium impact on confidentiality and integrity, with no impact on availability.

A fix is available in Tandoor Recipes version 2.6.4 that addresses this vulnerability by properly sanitizing input and disallowing unsafe <style> tag injection. Users should upgrade to version 2.6.4 or later. Until upgraded, clients consuming the instructions_markdown field should implement additional sanitization or avoid rendering this content as HTML without proper filtering. There is no vendor advisory explicitly stating no action is required, so patching or applying additional sanitization is recommended.