DB-GPT 0.8.2 Endpoint Error Grants Unauthenticated Root Code Execution
A flaw in DB-GPT 0.8.2 allows attackers to run code as root without login, exposing containers and databases to full compromise.
Key points
- CVE-2026-77932 carries a CVSS score of 9.3, rated critical by the NVD.
- The vulnerability affects eosphoros-ai DB-GPT versions up to and including 0.8.2.
- Attackers can execute arbitrary Python or bash commands as root via the API.
An unauthenticated remote code execution flaw in DB-GPT 0.8.2 allows attackers to gain full root access within the webserver container. According to the National Vulnerability Database, this critical issue stems from a misconfigured endpoint that incorrectly assigns administrative privileges to any caller, enabling arbitrary command execution without prior authentication.
How it unfolded
- The NVD published CVE-2026-77932, identifying a critical flaw in DB-GPT through version 0.8.2.
- The vulnerability involves the /api/v1/chat/react-agent endpoint relying on a mock user function.
- This function grants every caller admin role, bypassing authentication checks entirely.
Who is affected
Organizations running eosphoros-ai DB-GPT versions up to 0.8.2 are vulnerable to this attack. The flaw exposes the webserver container to unauthorized control, allowing attackers to run unsandboxed tools. According to the NVD record, this includes the ability to access bundled databases using default credentials, significantly expanding the scope of potential data exposure and system compromise for affected deployments.
The fix
No patch or update has been confirmed for this vulnerability yet. Users of DB-GPT 0.8.2 and earlier versions should assume their systems remain exposed until the vendor releases a remediation. Until then, the risk of unauthenticated root access remains active for any instance accessible to potential attackers, requiring immediate attention from security teams managing these environments.
What made it possible
This incident highlights the danger of improper privilege management in API endpoints. When developers use mock functions for testing that persist in production, they often bypass essential authentication layers. Teams usually verify that all endpoints enforce strict identity checks before granting elevated permissions. Check whether your APIs rely on hardcoded roles or simulated user data, as these can create backdoors that allow attackers to escalate privileges without valid credentials, undermining the entire security model of the application.
What to do and how to stay safe: DB-GPT
- Restrict network access to the DB-GPT API endpoints to trusted internal networks only.
- Monitor server logs for unusual execution patterns or unauthorized administrative actions.
- Verify that default database credentials are changed immediately if the service is exposed.
- Isolate affected containers to prevent lateral movement if a breach is suspected.
Step-by-step guide: What Is Virtual Patching: How It Works Without Code Changes
General security guidance from the RedTeam Tribune newsroom. It is not confirmed advice from the organisations named in this story.
Frequently asked questions
What is the CVSS score for CVE-2026-77932?
The National Vulnerability Database rates CVE-2026-77932 as critical with a CVSS score of 9.3.
Which versions of DB-GPT are affected by this flaw?
All versions of eosphoros-ai DB-GPT up to and including 0.8.2 are affected by this vulnerability.
Can attackers access the database through this exploit?
Yes, attackers can reach the bundled database using default credentials via the compromised container.
