DB-GPT 0.8.2 Vulnerability Grants Unauthenticated Root Access to Containers

DB-GPT versions through 0.8.2 contain a critical security vulnerability that exposes organizations to severe compromise risks. The flaw, tracked as CVE-2026-77932, carries a CVSS score of 9.3, indicating a high severity threat. This issue stems from a specific endpoint that improperly handles user authentication, effectively granting administrative privileges to any caller. Systems running these versions are directly at risk if the application is accessible to potential attackers.

How it unfolded

• The National Vulnerability Database recorded CVE-2026-77932 as affecting DB-GPT through version 0.8.2
• The vulnerability was identified within the /api/v1/chat/react-agent endpoint of the software
• The flaw relies on a mock get_user_from_headers function that grants admin roles to all callers
• Unauthenticated attackers can exploit this to trigger code execution without prior login credentials

Who is affected

eosphoros-ai DB-GPT installations running version 0.8.2 or earlier are affected by this vulnerability. The flaw allows unauthenticated remote code execution because the application incorrectly assigns admin roles to every user interacting with the specific endpoint. This means any entity that can reach the service can potentially take control of the container. The risk is particularly high for deployments that are internet-facing or have minimal network segmentation protecting the database connections.

Our take on the response

The severity of this flaw is underscored by the ability to run arbitrary code as root. The mock function essentially bypasses all authentication mechanisms, treating every request as if it came from an administrator. This design error exposes the unsandboxed code interpreter tools to direct manipulation. Organizations should treat this as an urgent priority, as the path from initial access to full system control is short and requires no valid credentials.

Who needs to act first

Teams should immediately inventory all DB-GPT instances to identify those running vulnerable versions. Prioritize remediation for systems exposed to untrusted networks or the internet, as these face the highest risk of exploitation. Verify whether network access controls can restrict connectivity to the specific endpoint while a solution is developed. Ensure that backup procedures are current and tested, providing a recovery path if compromise occurs. Monitor logs for unusual activity that might indicate unauthorized code execution or shell access attempts.

Background: Network access control

Network access control is a security mechanism that checks devices before allowing them onto your network. It verifies identity and compliance with security policies. This system blocks unauthorized or non-compliant devices, protecting your infrastructure from internal threats and limiting the spread of malware.

Read the full guide: Network Access Control Definition: How Policies Block Unauthorized Devices

What to do and how to stay safe: DB-GPT

• Scan your environment to locate all instances of DB-GPT and verify their current version numbers
• Implement network segmentation to restrict access to the vulnerable endpoint from untrusted sources
• Monitor application logs for signs of unauthorized code execution or unexpected administrative actions
• Prepare to apply updates immediately once the vendor releases a patched version of the software

Step-by-step guide: Vulnerability Prioritization: How to Rank Risks Without Wasting Time

General security guidance from the Patch Observer newsroom. It is not confirmed advice from the organisations named in this story.

Frequently asked questions

What is the CVSS score for this vulnerability?

The CVSS score for CVE-2026-77932 is 9.3, which is classified as critical.

Which versions of DB-GPT are affected?

DB-GPT versions through 0.8.2 are affected by this vulnerability.

Is there a patch available for this issue?

No fix has been confirmed yet according to the source material.