Impact
Unrestricted execution of arbitrary Python or bash code as root within the web server container is possible because the /api/v1/chat/react-agent endpoint relies on a mock function that grants every caller an admin role. Prompt templates designed for the ReAct agent can trigger the unsandboxed code_interpreter or shell_interpreter tools, allowing an unauthenticated attacker to run arbitrary commands with full container privileges, including connecting to the bundled database using its default credentials. This enables compromise of the entire application environment and any data stored in the database.
Affected Systems
The vulnerability affects the eosphoros-ai DB-GPT application, versions through 0.8.2. All deployments of DB-GPT at or below 0.8.2 are susceptible; newer releases are not referenced as affected.
Risk and Exploitability
The CVSS score of 9.3 reflects high severity. The EPSS score is not available, and the vulnerability is not listed in CISA's KEV catalog, but the lack of authentication combined with the ability to run arbitrary code makes exploitation highly likely in a real‑world setting. The attack vector is inferred to be an unauthenticated HTTP request to the /api/v1/chat/react-agent endpoint, as the mock authentication bypass satisfies the tool access controls.
OpenCVE Enrichment