ZeptoClaw’s generic webhook channel accepts caller‑supplied identity fields (sender and chat_id) from the request body and performs authorization checks on these untrusted values. Because authentication on the webhook is optional and defaults to disabled, an attacker who can reach the POST /webhook endpoint can send a request with a spoofed allowlisted sender and any arbitrary chat_id, allowing the attacker to inject high‑risk messages as another user or to route a session to a different chat entity. The flaw is a classic example of missing authentication and use of insufficient primary authentication, which can lead to confidentiality and integrity violations.

The affected vendor is qhkm, product ZeptoClaw, a personal AI assistant. Versions prior to 0.7.6 are vulnerable. The fix was released in the 0.7.6 release.

The CVSS score of 8.2 indicates a high severity vulnerability, but the EPSS score of less than 1% suggests a low probability of exploitation in the wild. The vulnerability is not listed in CISA’s KEV catalog. Exploitation requires network reachability to the ZeptoClaw instance and does not require privileged access; the attacker only needs to craft a POST request with spoofed identity data. If the webhook endpoint is exposed, attackers could weaponize the flaw to spoof messages or hijack chat flows. Defensive action is strongly recommended.