Capgo CVE-2026-56331

The /private/accept_invitation endpoint must be network-accessible, which is the default posture for both cloud-hosted and self-hosted Capgo deployments. No authentication is required - the CVSS 4.0 vector confirms PR:N (no privileges required) - and the attacker needs only the ability to send HTTP requests to the endpoint using a publicly known path. The actual volume of information disclosed in HTTP 500 responses varies by server configuration: deployments running with debug-level error verbosity or without a global exception handler will leak significantly more detail than production-hardened instances. Exploitation is strictly limited to information disclosure; this vulnerability alone does not enable data modification or remote code execution.

The CVSS 4.0 base vector (AV:N/AC:L/AT:N/PR:N/UI:N/VC:L/VI:N/VA:N/SC:N/SI:N/SA:N) reflects a low-complexity, unauthenticated, network-reachable information disclosure with limited confidentiality impact and no integrity or availability consequence. This vulnerability is not listed in CISA KEV and no public exploit code has been identified at time of analysis; EPSS data was not provided. The overall risk is moderate for self-hosted or cloud-hosted Capgo deployments exposed to the internet: the vulnerability is not a standalone critical threat but functions as a reconnaissance enabler - leaked server internals can inform subsequent exploitation attempts against the platform. Deployments with verbose error reporting or debug-mode configurations will expose significantly more sensitive detail than hardened instances, meaning actual confidentiality impact varies by deployment posture. The absence of any integrity or availability impact limits this to a supporting role in a multi-stage attack.

An unauthenticated attacker sends repeated HTTP POST requests to the publicly accessible /private/accept_invitation endpoint with intentionally malformed magic_invite_string payloads. Each request triggers an unhandled server exception that returns an HTTP 500 response potentially containing stack traces, framework version details, or internal path information, which the attacker harvests to profile the server environment for use in follow-on attacks. No public exploit code has been identified at time of analysis.