/

CVE-2026-72868

Severity:

CRITICAL RISK

(9.9)

OS Command Injection (CWE-78) in Dokploy Platform

CVE-2026-72868

Received

Vulnerability type

Category:

OS Command Injection

CWE:

CWE-78, CWE-862

Impact:

Remote Code Execution and Potential Host Compromise

Severity:

CRITICAL RISK

(9.9)

CVSS version:

3.1

Catch this CVE while code is being written

Detect vulnerable code directly in pull requests, so CVEs are fixed before merge.

Summary

A command injection vulnerability in Dokploy allows a low-privileged organization member to influence arguments passed to an rclone ls command through the destination connection test feature. Under deployments where the Dokploy container can access the host Docker socket, successful exploitation may enable arbitrary command execution with a path toward host-level compromise.

Affected context

The issue occurs in Dokploy's destination connection testing workflow when destination fields are incorporated into a shell command executed by the server. It is relevant where an organization member has permission to create destinations, attacker-controlled values reach the command construction logic, and the Dokploy container retains access to a Docker socket or other highly privileged runtime resources.

Exploit preconditions

  • Attack surface: The destination creation and connection-testing API
  • Required access: A low-privileged organization membership with permission to create destinations
  • Trigger point: Processing a destination test request that invokes the server-side rclone ls command
  • Why exploitation is possible: User-controlled destination fields are interpolated into a shell command without sufficient argument isolation, allowing command syntax to alter the intended execution context
  • Privilege dependency: Impact is increased when the Dokploy container can communicate with the host Docker daemon

Why this happens

The underlying bug is the construction of a shell command by interpolating attacker-influenced values into a string passed to a command execution API. This creates an OS command injection condition because data intended as command arguments can be interpreted as shell syntax. The issue maps primarily to CWE-78: Improper Neutralization of Special Elements used in an OS Command, while the authorization path also relates to CWE-862: Missing Authorization when a low-privileged member can reach a sensitive operation without adequate privilege separation.

Potential impact

  • An attacker may execute arbitrary commands inside the Dokploy container.
  • The attacker could access application secrets, mounted files, or service credentials available to the container.
  • A compromised container may be used to interact with the host Docker daemon.
  • The attacker could create, modify, or destroy containers and workloads.
  • The vulnerability could enable service disruption, data exposure, or lateral movement within connected environments.

General mitigation guidance

Mitigation generally focuses on removing shell interpretation from command execution and reducing the privileges available to the application.

Common defensive patterns include:

  • Use argument-array process APIs that execute the intended binary directly, rather than constructing shell command strings.
  • Apply strict allowlists and type validation to provider, region, endpoint, bucket, and credential-related fields.
  • Treat all request values as untrusted, including values supplied by authenticated organization members.
  • Enforce explicit authorization checks for destination creation and connection testing, with least-privilege role boundaries.
  • Avoid exposing the host Docker socket to application containers; use a restricted mediation service when container management is required.
  • Run the service with minimal filesystem, network, and process privileges and isolate credentials from general application workloads.

How teams detect issues like this

Detection efforts can include:

  • Static analysis rules for shell command construction using string interpolation or concatenation with request-controlled values.
  • Code review checks for child_process.exec and similar APIs receiving dynamically assembled command strings.
  • Tests that verify special characters in destination fields remain data and cannot change process execution behavior.
  • Authorization reviews for mutations accessible to organization members with limited privileges.
  • Runtime monitoring for unexpected child processes, unusual rclone arguments, container-management activity, or Docker socket access.
  • Policy enforcement that blocks host Docker socket mounts unless explicitly justified and isolated.

Automated code-scanning and security review tools may help identify command-execution and authorization patterns early.

Sources

SourceLink
github.comgithub.com/Dokploy/dokploy/commit/eeb6e7b8ea88e4b4b1fac8460755464100516ac9
github.comgithub.com/Dokploy/dokploy/pull/4873
github.comgithub.com/Dokploy/dokploy/releases/tag/v0.29.13
github.comgithub.com/Dokploy/dokploy/security/advisories/GHSA-f6x8-vfwh-8hjr

Frequently asked

questions

It can be reachable through Dokploy's application API when an attacker has an organization account with access to the destination creation and testing workflow. The exact exposure depends on network reachability and deployment authorization settings.

The described path requires a low-privileged organization member who can invoke the destination creation mutation. Anonymous access is not implied.

Avoid shell-based command construction and execute external programs through APIs that pass arguments separately. Combine this with strict input validation and narrowly scoped authorization checks.

Treat authenticated user input as untrusted, review every process-spawning call, apply least privilege to organization roles and containers, and avoid granting application containers direct access to host control sockets.

Find out what's already
exploitable in your codebase.