Apache Camel’s camel-mail component can copy every MIME header from an incoming multipart message into the Camel Exchange when headersInline is set to true. The names of those headers originate from the payload and are added without applying any header filter strategy. If the attacker can influence the MIME message, they can inject a header whose name falls into Camel’s internal namespace. Camel components such as the camel-sql producer read such internal headers to override configuration or supply a SQL statement. Therefore, an injected header could redirect downstream processors or let the route execute unintended logic, which may result in data modification, exfiltration, or other amplification of the route’s behavior.

Vulnerable versions of Apache Camel are all releases from 2.17.0 up to 4.14.9, from 4.15.0 up to 4.18.4, and from 4.19.0 up to 4.22.0. The issue is present in the camel-mail component across these lines. Versions prior to 4.14.9, 4.18.4, and 4.22.0 do not contain the fix.

The vulnerability is an instance of improper input validation (CWE‑20). Exploitation requires a crafted MIME multipart message sent to a camel‑mail endpoint with headersInline enabled; the attack surface is therefore an inbound channel such as email or HTTP that accepts MIME data. The EPSS score of less than 1% indicates a low likelihood of exploitation, and the issue is not listed in the CISA KEV catalog. The potential impact is high if the route consumes control headers, as an attacker could redirect downstream processing or execute unintended logic. The CVSS score of 6.5 signifies a moderate severity.