CVE-2025-47913

A flaw in golang.org/x/crypto/ssh/agent causes the SSH agent client to panic when a peer responds with the generic SSH_AGENT_SUCCESS (0x06) message to requests expecting typed replies (e.g., List, Sign). The unmarshal layer produces an unexpected message type, which the client code does not handle, leading to panic("unreachable") or a nil-pointer dereference. A malicious agent or forwarded connection can exploit this to terminate the client process.

This vulnerability was marked as Important because it allows any malicious or misbehaving SSH agent to force a crash in the client process using a single valid protocol byte. The panic occurs before the client has a chance to validate message structure or recover, which means an attacker controlling—or intercepting—SSH agent traffic can reliably terminate processes that rely on agent interactions. In environments where SSH agents operate over forwarded sockets, shared workspaces, or CI/CD runners, this turns into a reliable, unauthenticated remote denial of service against critical automation or developer tooling. The flaw also stems from unsafe assumptions in the unmarshalling logic, where unexpected but protocol-legal message types drop into “unreachable” code paths instead of being handled gracefully—making it a design-level reliability break rather than a simple error-handling bug. For this reason, it is rated as an important availability-impacting vulnerability rather than a moderate issue.

The CVSS score(s) listed for this vulnerability do not reflect the associated product's status, and are included for informational purposes to better understand the severity of this vulnerability.

No mitigation is currently available that meets Red Hat Product Security’s standards for usability, deployment, applicability, or stability.