Kestra renders user‑supplied metadata from flow YAML files through a Markdown component with HTML enabled. The rendered content is directly injected into the page via Vue's v‑html directive without sanitization, allowing a flow author to embed malicious JavaScript. When any user views or interacts with the affected flow, the script executes in that user's browser, potentially stealing credentials or executing further actions.

This vulnerability exists in all releases of Kestra up to and including version 1.3.3 issued by kestra‑io, affecting the Markdown.vue renderer used for flow metadata fields such as description, inputs[].displayName, and inputs[].description.

The flaw carries a CVSS score of 7.3, indicating high severity, while its EPSS score is below 1 % and it is not listed in the CISA KEV catalog, pointing to low probability of widespread exploitation. Nonetheless, the attack does not require any interaction from the victim beyond opening the flow, as the payload is stored and executed automatically. An attacker who can author or modify flows can leverage this to run arbitrary JavaScript in the context of any viewer’s session.