Parse Dashboard CVE-2026-27609

Exploitation requires that the target Parse Dashboard instance has the optional 'agent' configuration block enabled (dashboards without an agent config are entirely unaffected), AND that a victim with a currently authenticated dashboard session is tricked into visiting attacker-controlled web content (UI:P). The attacker needs no credentials of their own (PR:N) and no access to the dashboard host, but cannot succeed against an unauthenticated victim, a logged-out session, or a deployment lacking the agent feature. These factors - non-default agent configuration, an active authenticated session, and required user interaction - meaningfully limit realistic exploitation despite the network attack vector.

Signals are mixed and point to elevated-but-not-urgent risk. The vendor CVSS 4.0 base score is 8.3 (High), driven by AV:N, AC:L, PR:N, UI:P, with VI:H and SI:H (high integrity impact on both the vulnerable and subsequent systems) and no confidentiality or availability impact - consistent with a CSRF that performs unauthorized state changes but does not directly read data. Against that, EPSS is only 0.02% (4th percentile), there is no CISA KEV listing, and no public proof-of-concept is referenced, so real-world exploitation pressure is currently low. The key limiting factors temper the high CVSS: exploitation requires an already-authenticated victim to be socially engineered into visiting attacker-controlled content (UI:P), and only deployments that actually enable the optional 'agent' configuration block are affected. Note a nuance worth verifying: CVSS PR:N reflects that the attacker holds no credentials, but successful exploitation still depends on the victim holding a valid authenticated dashboard session - this is the standard CSRF modeling, not unauthenticated direct access.

An attacker hosts a malicious web page containing an auto-submitting form or script targeting POST /apps/:appId/agent and lures a logged-in Parse Dashboard operator to visit it. The victim's browser automatically attaches the dashboard session, so the forged request reaches the agent endpoint and performs attacker-chosen agent actions under the victim's identity. No public PoC is referenced, and the attack hinges on the low-complexity network vector (AC:L) combined with the required victim interaction (UI:P).