A flaw in the eosphoros‑ai db‑gpt component exposes the /api/v1/editor/ endpoint to SQL injection attacks. The vulnerability is exploitable from a remote attacker who can send crafted input that is interpreted directly by the database. Successful exploitation could allow the attacker to read, modify, or delete data in the database, and under certain conditions, may provide a foothold for further compromise or remote code execution. The CVSS score of 6.9 indicates that the impact is significant but does not guarantee full system takeover. The publicly available exploit demonstrates that the flaw is already in the wild and can be used without additional privileges.

The vulnerability affects eosphoros‑ai db‑gpt versions up to 0.7.5. No single revision beyond 0.7.5 is known to be impacted, and there is no information indicating that any earlier or later branches are remedied. The attack is confined to the exposed web API, so any deployment of the affected version that allows remote access to /api/v1/editor/ is susceptible.

The reported CVSS score of 6.9 reflects a medium‑high risk, and the EPSS score is not available. The vulnerability is not listed in CISA’s KEV catalog, but an exploitation tool has already been published, suggesting that attackers may target vulnerable installations. The attack vector is remote, via HTTP requests to the editor API, and no privileged local execution is required. Given the absence of a vendor response and the lack of an available patch, the risk remains high until the affected component is removed or secured.