Untrusted Code Execution in DB-GPT HuggingFace Adapter

A weakness has been identified in eosphoros-ai DB-GPT up to 0.8.2. Impacted is the function exec_module of the file hf_adapter.py of the component HuggingFace Adapter. Executing a manipulation of the argument trust_remote_code can lead to inclusion of functionality from untrusted control sphere. The attack can be launched remotely. The exploit has been made available to the public and could be used for attacks. The project was informed of the problem early through an issue report but has not responded yet.

Casky was already ahead

This CVE exploits attack patterns that Casky's 447matched skills already investigate — long before this vulnerability was disclosed. Claude's reasoning model maps these techniques to MITRE ATT&CK, so practitioners who ran these skills have already seen the threat behaviour in their findings.

Analysis

CVE-2026-123019 represents a critical supply chain and code execution vulnerability in eosphoros-ai DB-GPT versions up to 0.8.2, specifically within the HuggingFace Adapter component. By manipulating the trust_remote_code argument in the exec_module function of hf_adapter.py, attackers can inject and execute arbitrary code from untrusted sources. This weakness is particularly dangerous because it affects AI/ML infrastructure that organizations increasingly rely on for production workloads. The vulnerability requires no authentication and can be exploited remotely, making it accessible to threat actors with minimal prerequisites. Organizations using DB-GPT for data processing, model inference, or integration with HuggingFace model repositories face direct risks of system compromise, data exfiltration, and lateral movement within their environments.

Casky's 447 matched security skills enable practitioners to detect the attack patterns underlying this vulnerability across the reconnaissance, execution, and persistence phases (TA0001, TA0043, TA0002). Using Claude AI with extended reasoning, practitioners can identify telemetry signatures such as unexpected remote code execution contexts, suspicious HuggingFace model repository interactions, and anomalous trust_remote_code parameter configurations. The platform helps detect both the initial exploitation attempt—when attackers manipulate trust_remote_code to load malicious model code—and post-exploitation activity like data staging and command execution. Casky findings would highlight unusual process spawning from Python environments, unexpected network connections to model repositories, and attempts to execute code outside normal application workflows, enabling defenders to correlate these indicators with the specific MITRE techniques leveraged in active exploitation.

Live Threat IntelligenceComing soon

Casky Risk Score

—

EPSS Probability

—

Shodan Exposure

—

GreyNoise Signal

—

Composite risk scoring from EPSS, CISA KEV, Shodan, and GreyNoise — 21 security APIs correlated into a single Casky Risk Score. Coming in Casky Pro. Join early access →

447 Casky skills cover this attack pattern

These skills use Claude AI's reasoning model to surface findings in the same attack categories as CVE-2026-123019.

Abuse of Public-Facing API: Mobile API Abuse

mobile security · low

Run

abusing-dpapi-for-credential-access

red teaming · high

Run

abusing-shadow-credentials-for-privesc

red teaming · high

Run

MITRE ATT&CK Techniques

TA0001 TA0043 TA0002

Investigate the attack patterns behind CVE-2026-123019

Casky has 447 skills that investigate the attack patterns behind CVE-2026-123019. Run one and get CVSS-scored findings in 3 minutes.

Run the skill that detects this →

X Instagram LinkedIn