The vulnerability in Aix‑DB 1.2.4 permits authenticated non‑administrative users to execute arbitrary operating‑system commands via the ChatBI common agent's ungated deepagents execute functionality. By sending specially crafted chat messages to the POST /dify/get_answer endpoint or uploading documents that contain malicious text, an attacker can trigger the exec function, leading to remote code execution. This flaw exposes datasource credentials and model API keys, enabling full compromise of the underlying system.

The vulnerability affects the Apconw Aix‑DB product, specifically version 1.2.4. No other versions are currently known to be impacted, and no broader vendor or product list is indicated. Reducing exposure requires restricting access to this version or applying the appropriate fix.

The CVSS score of 8.7 indicates high severity, while the EPSS score is currently unavailable – the actual exploitation probability remains unknown. The flaw is not listed in the CISA KEV catalog. Because authentication is required, only users with valid credentials can exploit the flaw, but the malicious payload can be triggered remotely over the network, making the attack vector remote yet authenticated. Without an immediate patch, attackers could leverage exposed credentials to mount further attacks.