External risk intelligence

DB-GPT Path Traversal Arbitrary File Write Vulnerability

CVE advisorySeverity: CRITICAL (CVSS 9.3)

CVE-2026-73034

The vulnerability resides in a file-upload endpoint of a web application platform. Web applications and their associated API endpoints are commonly deployed as internet-facing services, making this surface frequently reachable in standard web deployment scenarios.

Path Traversal

Halo Surface Signal: 4 out of 5 — likely to be public-facing.

External exposure likelihood

Horizon Alert

Summary of the vulnerability and why it matters

A critical vulnerability has been identified in DB-GPT that could allow attackers to execute arbitrary code on affected servers. This issue stems from a path traversal flaw in the file upload functionality, which, if exploited, could enable unauthorized writing of files to sensitive system locations.

  • Allows attackers to write files anywhere on a server.
  • Critical risk if unauthenticated remote code execution is possible.
  • Confirm relevance and assess exposure to this application.

Attack Path

How an attacker could exploit the issue

An attacker can exploit this vulnerability by sending a specially crafted request to the Python file-upload endpoint. This request manipulates the `user_id` HTTP header with directory traversal sequences, allowing the attacker to bypass intended file upload restrictions and write files to arbitrary server locations. This could lead to the execution of malicious code on the server.

  • No authentication or special access needed.
  • Malicious user ID header in file upload.
  • Arbitrary file write leading to code execution.

Live Threat

Current exploitation, exposure, and threat context

An unauthenticated path traversal vulnerability in DB-GPT could allow remote attackers to write arbitrary files to any server location. This may occur when a crafted multipart upload request targets the Python file-upload endpoint, enabling attackers to potentially execute arbitrary code by overwriting critical system files or scripts.

  • Arbitrary file write on server.
  • Via crafted HTTP header and file upload.
  • Leads to potential remote code execution.

Operational Fix

Recommended remediation, mitigation, and detection steps

System administrators and application owners are responsible for addressing this critical vulnerability in DB-GPT. The first practical step is to identify all instances of DB-GPT within your environment, determine their exposure and business criticality, and confirm the accountable owner for each instance before planning remediation.

  • Confirm ownership and assess exposure.
  • Verify system reachability and business impact.
  • Plan targeted remediation based on risk.

Supplementary metadata

Validate whether this threat affects your internet-facing exposure.

Halo Threat Intelligence helps prioritize remediation with Halo Surface Signal and H/A/L/O context. Start exposure validation with a free external attack surface trial.

Frequently asked questions

What is DB-GPT and what is it used for?

DB-GPT is an open-source framework designed for building data-driven applications using large language models. It acts as an integration platform that connects LLMs with databases and specialized tools, allowing developers to create agents that interact with private or public data to automate complex tasks and generate insights.

What is the path traversal flaw in CVE-2026-73034?

This vulnerability is classified as CWE-22, Improper Limitation of a Pathname to a Restricted Directory. In simple terms, the application fails to properly sanitize user input, allowing an attacker to use special file-path characters to 'escape' the intended upload folder. This allows them to write files to unauthorized, sensitive areas of the server filesystem.

How can an attacker trigger this vulnerability?

An attacker triggers this by sending a specially crafted multipart HTTP request to the file-upload endpoint. By injecting directory traversal sequences—such as '../'—into the user_id HTTP header, they force the server to save an uploaded file to a directory of their choosing. Requests that do not include these specific, malicious traversal sequences in that header do not trigger the flaw.

Do I need to worry if my DB-GPT instance is internal?

Halo Surface Signal indicates that because this vulnerability exists in a web-based file-upload endpoint, it is typically deployed as an internet-facing service, which significantly increases risk. Even if your instance is currently internal, you should prioritize reviewing its network reachability to ensure it is not inadvertently accessible from broader, less-trusted network segments.

What are the first steps to secure my environment?

Start by identifying all running instances of DB-GPT across your infrastructure to understand your potential footprint. Once you have an inventory, assess the business criticality of each instance and verify its specific network exposure. Coordinate with the relevant system owners to ensure they are aware of the risk and ready to implement the necessary updates or mitigation measures.

References