Parse Dashboard
Monthly
Privilege escalation via cache key collision in Parse Dashboard (npm parse-dashboard) versions 7.3.0-alpha.42 through 9.0.0-alpha.7 allows an authenticated read-only user to obtain the full master key. The flaw lives in ConfigKeyCache, which reuses one cache entry for both the master key and the read-only master key when those keys are function-typed; under a timing race a read-only session can read back the cached full master key (or a regular user the read-only key). Exploitation requires an authenticated dashboard session and a non-default configuration (function-typed keys plus the agent block); no public exploit identified at time of analysis and EPSS is low (0.05%), but the GitHub Security Advisory and patch commit confirm the issue.
Cross-site request forgery in Parse Dashboard (versions 7.3.0-alpha.42 through 9.0.0-alpha.7) lets a remote attacker abuse the unprotected AI Agent endpoint (POST /apps/:appId/agent), which lacks CSRF protection and a session-bound token. By luring an authenticated dashboard operator to a malicious web page, the attacker forces agent requests to execute under the victim's session, driving state-changing operations against managed Parse Server apps. There is no public exploit identified at time of analysis, EPSS risk is negligible (0.02%), and the issue is fixed in 9.0.0-alpha.8.
Broken authorization in Parse Dashboard's AI Agent endpoint (POST /apps/:appId/agent) lets any authenticated user reach apps they are not scoped to and lets read-only users escalate to full read-write control. Present in versions 7.3.0-alpha.42 through 9.0.0-alpha.7, the endpoint omits per-app authorization checks and hands read-only users the full master key, so an attacker can swap the app ID in the URL to act against other tenants and supply write permissions to perform write/delete operations. No public exploit is identified at time of analysis, and the EPSS probability is very low (0.03%), but it is trivially exploitable by any logged-in low-privilege user where the agent feature is enabled.
Unauthenticated arbitrary database read/write in Parse Dashboard's AI Agent endpoint (POST /apps/:appId/agent) lets remote attackers operate against any connected Parse Server using the master key. Affected builds span the 7.3.0-alpha.42 through 9.0.0-alpha.7 pre-releases, but only deployments that have explicitly enabled the opt-in agent feature are exposed. No public exploit identified at time of analysis, and EPSS is very low (0.04%), but the chained authentication/CSRF/authorization gaps make exploitation trivial where the feature is configured.
Privilege escalation via cache key collision in Parse Dashboard (npm parse-dashboard) versions 7.3.0-alpha.42 through 9.0.0-alpha.7 allows an authenticated read-only user to obtain the full master key. The flaw lives in ConfigKeyCache, which reuses one cache entry for both the master key and the read-only master key when those keys are function-typed; under a timing race a read-only session can read back the cached full master key (or a regular user the read-only key). Exploitation requires an authenticated dashboard session and a non-default configuration (function-typed keys plus the agent block); no public exploit identified at time of analysis and EPSS is low (0.05%), but the GitHub Security Advisory and patch commit confirm the issue.
Cross-site request forgery in Parse Dashboard (versions 7.3.0-alpha.42 through 9.0.0-alpha.7) lets a remote attacker abuse the unprotected AI Agent endpoint (POST /apps/:appId/agent), which lacks CSRF protection and a session-bound token. By luring an authenticated dashboard operator to a malicious web page, the attacker forces agent requests to execute under the victim's session, driving state-changing operations against managed Parse Server apps. There is no public exploit identified at time of analysis, EPSS risk is negligible (0.02%), and the issue is fixed in 9.0.0-alpha.8.
Broken authorization in Parse Dashboard's AI Agent endpoint (POST /apps/:appId/agent) lets any authenticated user reach apps they are not scoped to and lets read-only users escalate to full read-write control. Present in versions 7.3.0-alpha.42 through 9.0.0-alpha.7, the endpoint omits per-app authorization checks and hands read-only users the full master key, so an attacker can swap the app ID in the URL to act against other tenants and supply write permissions to perform write/delete operations. No public exploit is identified at time of analysis, and the EPSS probability is very low (0.03%), but it is trivially exploitable by any logged-in low-privilege user where the agent feature is enabled.
Unauthenticated arbitrary database read/write in Parse Dashboard's AI Agent endpoint (POST /apps/:appId/agent) lets remote attackers operate against any connected Parse Server using the master key. Affected builds span the 7.3.0-alpha.42 through 9.0.0-alpha.7 pre-releases, but only deployments that have explicitly enabled the opt-in agent feature are exposed. No public exploit identified at time of analysis, and EPSS is very low (0.04%), but the chained authentication/CSRF/authorization gaps make exploitation trivial where the feature is configured.