CVE-2026-101008: aaPanel BaoTa: command injection in file merge handler

CRITICALPoC AVAILABLECISA: ATTEND
Published September 28, 2026
CISO Take

aaPanel (BaoTa) up to 11.8.0 has an OS command injection in the file merge handler (merge_split_file in files.py), reachable remotely through the split_file_path argument. The flaw needs high privileges, so it is mainly a post-authentication escalation or insider risk. Even so, a public exploit write-up is linked in the references, the vendor has not responded to the disclosure, and no patched version exists. EPSS is low (0.023, top 17%), it is not in CISA KEV, and CISA SSVC rates it ATTEND. Note that the panel is a general web hosting control panel, not a tracked AI package, and the 'ml_ui' category looks like a misclassification, so it matters only where it hosts ML workloads or notebooks. Restrict panel access to trusted networks or a VPN, audit and rotate admin credentials, remove the panel from the internet, and monitor for unusual shell activity from the panel's Python process until the vendor ships a fix.

Sources:NVDEPSSCISA SSVCATLASvuldb.com

What is the risk?

Rated critical (CVSS 9.1), but the high-privilege requirement (PR:H) and the low EPSS (top 17%) reduce practical likelihood. Aggravating factors: a public exploit gist is referenced (the structured 'public exploit' flag says No, a discrepancy worth verifying), the vendor is unresponsive, there is no known patch, and the scope is changed (S:C), so a compromise can reach the host. The AI relevance is indirect and applies only if the panel manages servers that run ML workloads. Not in KEV, no Nuclei template, and no dependents data is available.

How does the attack unfold?

Initial Access
Attacker authenticates to an exposed aaPanel instance using stolen or weak admin credentials.
AML.T0049
Exploitation
A crafted split_file_path with shell metacharacters is sent to the file merge handler, injecting OS commands.
AML.T0050
Impact
Arbitrary code runs on the host, exposing co-hosted ML models, datasets and credentials and allowing lateral movement.

What systems are affected?

PackageEcosystemVulnerable RangePatched
BaoTa——No patch

Do you use BaoTa? You're affected.

How severe is it?

CVSS 3.1
9.1 / 10
EPSS
2.3%
chance of exploitation in 30 days
Higher than 84% of all CVEs
Exploitation Status
Exploit Available
Exploitation: MEDIUM
Sophistication
Trivial
Exploitation Confidence
medium
○CISA SSVC: Public PoC
○Public PoC indexed (trickest/cve)
Composite signal derived from CISA KEV, VulnCheck KEV, CISA SSVC, EPSS, Metasploit, Exploit-DB, trickest/cve, Nuclei templates, and inthewild.io exploitation reports.

What is the attack surface?

AVACPRUISCIA
AVNetwork
ACLow
PRHigh
UINone
SChanged
CHigh
IHigh
AHigh

What should I do?

1 step
  1. 1) No patched version is published, so check the vendor's channels for releases after 11.8.0 and apply them as soon as they exist. 2) Do not expose the panel to the internet. Restrict it by firewall, IP allow-list or VPN, and change the default port. 3) Enforce strong, unique admin credentials with MFA where available, and review admin accounts and sessions. 4) Monitor for requests to the file merge endpoint with unusual split_file_path values (shell metacharacters such as ;, |, $(), backticks). Alert on child shells spawned by the panel's Python process. 5) Review the public exploit write-up to build detections. 6) Isolate ML workloads from the panel host, and rotate secrets stored on it if compromise is suspected. 7) Consider migrating to a maintained alternative if the vendor stays unresponsive.

What does CISA's SSVC say?

DecisionAttend
Exploitationpoc
AutomatableNo
Technical Impacttotal

Source: CISA Vulnrichment (SSVC v2.0). Decision based on the CISA Coordinator decision tree.

How is it classified?

Which compliance frameworks are affected?

This CVE is relevant to:

EU AI Act
Article 15-Accuracy, robustness and cybersecurity
NIST AI RMF
MANAGE 2.3-Responses to risks and unknown risks are in place

Frequently Asked Questions

What is CVE-2026-101008?

aaPanel (BaoTa) up to 11.8.0 has an OS command injection in the file merge handler (merge_split_file in files.py), reachable remotely through the split_file_path argument. The flaw needs high privileges, so it is mainly a post-authentication escalation or insider risk. Even so, a public exploit write-up is linked in the references, the vendor has not responded to the disclosure, and no patched version exists. EPSS is low (0.023, top 17%), it is not in CISA KEV, and CISA SSVC rates it ATTEND. Note that the panel is a general web hosting control panel, not a tracked AI package, and the 'ml_ui' category looks like a misclassification, so it matters only where it hosts ML workloads or notebooks. Restrict panel access to trusted networks or a VPN, audit and rotate admin credentials, remove the panel from the internet, and monitor for unusual shell activity from the panel's Python process until the vendor ships a fix.

Is CVE-2026-101008 actively exploited?

Proof-of-concept exploit code is publicly available for CVE-2026-101008, increasing the risk of exploitation.

How to fix CVE-2026-101008?

1) No patched version is published, so check the vendor's channels for releases after 11.8.0 and apply them as soon as they exist. 2) Do not expose the panel to the internet. Restrict it by firewall, IP allow-list or VPN, and change the default port. 3) Enforce strong, unique admin credentials with MFA where available, and review admin accounts and sessions. 4) Monitor for requests to the file merge endpoint with unusual split_file_path values (shell metacharacters such as ;, |, $(), backticks). Alert on child shells spawned by the panel's Python process. 5) Review the public exploit write-up to build detections. 6) Isolate ML workloads from the panel host, and rotate secrets stored on it if compromise is suspected. 7) Consider migrating to a maintained alternative if the vendor stays unresponsive.

What systems are affected by CVE-2026-101008?

This vulnerability affects the following AI/ML architecture patterns: Web hosting control panels hosting ML workloads, Self-hosted notebook and model-serving hosts.

What is the CVSS score for CVE-2026-101008?

CVE-2026-101008 has a CVSS v3.1 base score of 9.1 (CRITICAL). The EPSS exploitation probability is 2.34%.

What is the AI security impact?

Affected AI Architectures

Web hosting control panels hosting ML workloadsSelf-hosted notebook and model-serving hosts

MITRE ATLAS Techniques

AML.T0049Exploit Public-Facing Application
AML.T0050Command and Scripting Interpreter

Compliance Controls Affected

EU AI Act: Article 15
NIST AI RMF: MANAGE 2.3

What are the technical details?

Original Advisory

A vulnerability was found in aaPanel BaoTa up to 11.8.0. Impacted is the function merge_split_file of the file /www/server/panel/class/files.py of the component File Merge Handler. Performing a manipulation of the argument split_file_path results in command injection. The attack is possible to be carried out remotely. The exploit has been made public and could be used. The vendor was contacted early about this disclosure but did not respond in any way.

Exploitation Scenario

An attacker obtains admin credentials for an exposed aaPanel instance, through credential stuffing, phishing or a compromised insider account. They send a crafted request to the File Merge Handler with a split_file_path containing shell metacharacters, and the payload executes as the panel user. From the shell they search the host for Jupyter tokens, cloud keys, model files and datasets, then plant a persistent backdoor or move laterally to the GPU and training servers that share the host or network.

Weaknesses (CWE)

CWE-74 — Improper Neutralization of Special Elements in Output Used by a Downstream Component: The product constructs all or part of a command, data structure, or record using externally-influenced input from an upstream component, but it does not neutralize or incorrectly neutralizes special elements that could modify how it is parsed or interpreted when it is sent to a downstream component.

  • [Requirements] Programming languages and supporting technologies might be chosen which are not subject to these issues.
  • [Implementation] Utilize an appropriate mix of allowlist and denylist parsing to filter control-plane syntax from all input.

Source: MITRE CWE corpus.

CVSS Vector

CVSS:3.1/AV:N/AC:L/PR:H/UI:N/S:C/C:H/I:H/A:H/E:P/RL:X/RC:C

References

Timeline

Published
September 28, 2026
Last Modified
September 28, 2026
First Seen
September 28, 2026

Related Vulnerabilities