eosphoros-ai db-gpt Flow Import Endpoint import importlib.machinery.SourceFileLoader.exec_module कोड इंजेक्शन

The code injection vulnerability in db-gpt allows an attacker to execute arbitrary code on the server hosting the application. This can lead to complete system compromise, including data exfiltration, modification, and denial of service. Given the remote accessibility of the affected endpoint, the blast radius is significant, potentially impacting any system connected to the vulnerable db-gpt instance. The availability of a public exploit dramatically increases the likelihood of exploitation, making it a high-priority security concern. Successful exploitation could allow attackers to gain persistent access and move laterally within the network.

Organizations utilizing db-gpt version 0.7.5, particularly those exposing the Flow Import Endpoint to external networks or untrusted sources, are at significant risk. Environments with limited security monitoring or input validation practices are especially vulnerable.

• python / server:

import os
import subprocess

# Check for suspicious files in the flow import directory
import_dir = '/path/to/db-gpt/awel/flow'
for filename in os.listdir(import_dir):
    if filename.endswith('.py') and 'eval(' in open(os.path.join(import_dir, filename)).read():
        print(f'Suspicious file found: {filename}')

• linux / server:

# Monitor process activity for unexpected python executions
journalctl -f -u db-gpt | grep 'python' | grep 'eval'

1. 2026-03-02Disclosure

   disclosure

1. आरक्षित2026-03-01
2. प्रकाशित2026-03-02
3. EPSS अद्यतन2026-03-23

Due to the lack of a fixed version, immediate mitigation is crucial. Implement strict input validation on all data passed to the Flow Import Endpoint. Consider temporarily disabling the endpoint if possible. Employ a Web Application Firewall (WAF) with rules to block suspicious requests targeting the /api/v1/serve/awel/flow/import endpoint, specifically looking for unusual file imports or execution attempts. Monitor system logs for unusual process activity or file modifications. Regularly scan the system for unauthorized code execution.