An unauthenticated attacker can fire off arbitrary spacecraft commands through NASA’s mission console, according to new research from Cycode. The defects sit in AIT-GUI, the browser-based operator interface for the open-source AMMOS Instrument Toolkit developed at NASA/JPL.
The issues, tracked as GHSA-p9r8-2q67-fp86 and rated 9.4 on the CVSS v3.1 scale, affect versions 2.5.1 and earlier and are fixed in 2.5.2. A separate record, CVE-2026-60112 (9.3 on CVSS v4), published by VulnCheck in late July covers the same missing-authentication defect.
Instead of honoring the configured host, the AIT-GUI server hardcodes its listener to 0.0.0.0 on port 8080. Every state-changing route ships with no authentication and no cross-site request forgery protection. Since the routes accept application/x-www-form-urlencoded bodies, browsers treat them as simple requests, so a cross-origin POST from any page an operator visits reaches the server with no preflight.
An attacker who can reach the port can issue instrument and spacecraft commands via POST /cmd, execute server-side scripts through POST /script/run, and run sequences via POST /seq, including files outside the intended directory.
Cycode frames the exposure in operational terms: a single unauthenticated POST translates directly into issued instrument commands rather than a cosmetic page takeover.
Release 2.5.2 binds to the configured host, defaults to localhost, adds same-origin checks for mutating requests, and confines script and sequence routes to their roots. Neither advisory reports exploitation or names a mission using the software. The fix commit was co-authored with Anthropic’s Claude Opus 4.8, which Cycode says reflects how modern research will work: human analysts alongside AI-assisted code review.
