Executive brief
A security vulnerability exists in tgpt, a terminal-based AI chat tool, specifically within its update handling mechanism on Linux and macOS. An attacker with local access to the system can execute unauthorized commands by manipulating the update process. This could allow a malicious user to gain higher privileges or interfere with the tool's operations on a shared computer.
Technical details
A command injection vulnerability exists in the `helper.Update` function within `helper.go` of the tgpt Update Handler component. The flaw stems from improper neutralization of special elements used in a command (CWE-77), allowing a local attacker to inject and execute arbitrary system commands. The vulnerability affects versions up to 2.11.1 on Linux and macOS platforms. Exploitation requires local access with low privileges and no user interaction. While a public exploit has been disclosed, the vendor has reportedly not responded to the disclosure, and no official patch is currently confirmed.
Affected products
- aandrew-me tgpt up to 2.11.1
Timeline
- 2026-05-09: disclosed: Public disclosure of the vulnerability and exploit.
- 2026-05-09: advisory: CVE-2026-8210 published.