CVE-2026-129987: Dokploy Command Injection
Critical CVSS (9.9) reachable over network.
TL;DR - CVE-2026-129987 is a critical authenticated command-injection flaw in Dokploy versions before 0.31.1. - Users able to update database passwords may trigger operating-system command execution, commonly with root-level impact. - Upgrade to 0.31.1 or later and investigate suspicious password changes or unexpected child processes.
CVE-2026-129987 is a critical Dokploy vulnerability affecting versions before 0.31.1. The flaw allows an authenticated user authorized to update a database password to trigger operating-system command execution through Bash command substitution. Because Dokploy deployments commonly run with root privileges, organizations should treat this as a potential host-compromise issue.
What Happened?
Dokploy is a self-hostable platform-as-a-service product. Before version 0.31.1, database-password validation allowed backtick characters. Dokploy subsequently interpolated the supplied password into Bash command strings used by password-change procedures.
In Bash, backtick-delimited content performs command substitution. An authenticated user with permission to update a database password could therefore cause arbitrary operating-system commands to run before the intended database command executed. Dokploy deployments commonly run with root privileges, so the practical impact can include complete compromise of the Dokploy host and workloads it manages.
Analyst’s Take: Patch this as a host-compromise issue, not as a narrow database-password bug. The vulnerable workflow requires authentication, but command substitution can execute before the database command succeeds, and Dokploy commonly runs with root privileges.
For background on how security incidents may become legally reportable breaches, see when a security incident becomes a reportable data breach.
| Field | Value |
|---|---|
| CVE | CVE-2026-129987 |
| Product | Dokploy |
| CVSS base score | 9.9 Critical |
| CVSS vector | Not included in the supplied NVD result; confirm the exact vector on the NVD detail page |
| Attack vector | Network-facing Dokploy management workflow; exact CVSS vector should be confirmed |
| Authentication | Required |
| Privileges required | An authenticated account authorized to update a database password |
| Affected versions | Dokploy versions before 0.31.1 |
| Fixed version | Dokploy 0.31.1 |
| Patch available | Yes |
| CISA KEV status | Not listed when checked |
The affected password-change procedures cover MySQL, PostgreSQL, MariaDB, MongoDB, and Redis workflows. Exploitation does not require compromising one of those databases directly; the vulnerable Dokploy management operation is the execution path.
What’s the Root Cause?
The root cause is a shell-injection chain rather than a database authentication failure. Dokploy accepted backticks in attacker-controlled database-password input, placed that value into a Bash command string, and allowed Bash to interpret the backticks as command substitution.
This pattern combines unsafe input handling with unsafe command construction. Passwords are data, but the affected implementation treated them as part of shell syntax. Even if the resulting database command failed, command substitution could execute first. Application-level validation or database success was therefore not a reliable security boundary.
The impact is amplified by process privileges. Dokploy commonly runs as root, so commands launched through the vulnerable workflow may inherit root privileges. Defenders should treat exploitation as potential host compromise, not merely unauthorized database-password modification.
Technical Notes
The vulnerable execution pattern can be represented conceptually as:
# Unsafe pattern: attacker-controlled password is embedded in shell syntax
password="attacker-controlled-value"
bash -c "database-client --password '$password' ..."
Review the exact source implementation and patch in the Dokploy commit and pull request listed in the references. Do not assume that escaping one shell metacharacter is sufficient unless the project’s fixed implementation demonstrates a complete safe construction strategy.
Who Needs to Act?
Administrators running Dokploy versions earlier than 0.31.1 need to act. The available NVD description does not provide a narrower affected range, so the safe interpretation is that every Dokploy release before 0.31.1 is affected.
The impacted workflows include password-change procedures for:
- MySQL
- PostgreSQL
- MariaDB
- MongoDB
- Redis
Inventory every Dokploy installation, including isolated management nodes, development environments, and instances deployed through automation. A system may be exposed even if its managed databases are not externally reachable, because the vulnerable operation executes through Dokploy itself.
Identify users who can update database passwords. Those permissions may be held by platform engineers, application administrators, automation identities, or tenant-level users depending on the deployment model. Review authorization assignments rather than assuming that only global Dokploy administrators can reach the affected operation.
Why This CVSS Score?
The reported CVSS base score is 9.9 Critical. That severity is consistent with a network-reachable management interface, the need for authentication but not necessarily full administrative control, and the possibility of arbitrary operating-system command execution with high impact.
The supplied NVD result did not include the complete CVSS vector string. The individual metric values, such as attack complexity, user interaction, scope, confidentiality, integrity, and availability impacts, should not be presented as confirmed facts. Confirm the vector directly on the NVD detail page before using it for automated prioritization or compliance reporting.
Operationally, interpret the score alongside deployment context. Authentication lowers the exposure compared with an unauthenticated flaw, but it does not make the issue low risk. An attacker who obtains a qualifying Dokploy account, abuses an over-permissioned user, or compromises an administrative session may be able to convert that access into root-level execution on the platform host.
The likely impact is broad: an attacker may alter Dokploy configuration, access secrets, modify deployments, tamper with managed workloads, establish persistence, or use the host as a stepping stone into connected infrastructure. These consequences justify urgent patching even though CISA had not listed the CVE in KEV when checked.
Has It Been Exploited?
CVE-2026-129987 was not listed in the CISA Known Exploited Vulnerabilities catalog at the time of assessment. There is no CISA KEV date-added entry, remediation deadline, required-action statement, or ransomware-campaign flag for this CVE.
No verified CVE-specific public proof of concept was established from the supplied research. Search results involving other Dokploy advisories or unrelated vulnerabilities should not be treated as evidence of a PoC for this issue.
The current status is therefore:
| Question | Status |
|---|---|
| CISA KEV listing | No |
| Confirmed exploitation in the wild | Not established |
| Verified CVE-specific public PoC | Not established |
| Exploitability concern | High, because the command-substitution mechanism is straightforward and impact may be root-level |
Absence from KEV does not prove that exploitation has never occurred. It means only that CISA had not included the CVE in its catalog when checked. Organizations should not wait for public exploitation evidence before patching, particularly where Dokploy is internet-accessible or where many users can change database credentials.
Related Security Topics
Organizations reviewing Dokploy’s management-plane exposure may also want to understand what CASB means and how cloud access security brokers work. CASB controls are not a patch for CVE-2026-129987, but they can support broader visibility and policy enforcement across cloud services.
Detection Guidance
Start by identifying exposure. Enumerate Dokploy instances and determine their exact versions. Review whether any installation is older than 0.31.1, whether its management interface is reachable from untrusted networks, and which accounts can update database passwords.
Then review activity around database-password changes. Look for unexpected changes across MySQL, PostgreSQL, MariaDB, MongoDB, and Redis resources, especially when the operator, source address, time, or target database is unusual. A password-change event by itself is not proof of exploitation, but it establishes a useful investigation window.
Successful exploitation may create child processes, so correlate Dokploy activity with operating-system process telemetry. Look for shells or utilities spawned by the Dokploy service account, particularly around password-change requests. The following examples are hunting patterns, not confirmed CVE-specific indicators.
Technical Notes
A Linux process-audit query using ausearch might identify shell execution by a Dokploy service context:
sudo ausearch -m EXECVE -ts recent \
| grep -E '(/bin/(ba|z)?sh|/usr/bin/(curl|wget|nc|ncat|python|perl)|dokploy)'
If auditd is enabled, a more targeted rule can monitor executions by the account running Dokploy. Replace dokploy with the actual service account:
sudo auditctl -a always,exit -F arch=b64 \
-S execve -F euid=dokploy -k dokploy-child-processes
sudo ausearch -k dokploy-child-processes -i
A SIEM or EDR query should flag a Dokploy process spawning shells or network-capable utilities. A generic Sigma-style detection concept is:
title: Dokploy Service Spawns Shell Or Network Utility
logsource:
category: process_creation
detection:
parent:
ParentImage|endswith:
- "/dokploy"
child:
Image|endswith:
- "/sh"
- "/bash"
- "/zsh"
- "/curl"
- "/wget"
- "/nc"
- "/ncat"
- "/python"
- "/perl"
condition: parent and child
level: high
Review shell history where available, systemd journal entries, container runtime events, outbound connections, newly created users or SSH keys, modified deployment manifests, and unexpected changes to secrets. Logging may be incomplete, particularly if Dokploy runs in a container, so absence of a matching event does not rule out compromise.
Remediation and Patching
Upgrade every affected Dokploy installation to version 0.31.1 or later. The fixed version is explicitly identified in the NVD record and the project’s release references. After upgrading, verify the running version rather than relying solely on the package or container tag used by deployment automation.
A generic Git-based installation can be updated using the project’s documented deployment process. Do not substitute an unverified command for your environment. Where the installation is managed from a Git checkout, check the version target before applying the project’s normal update procedure:
git fetch --tags
git checkout v0.31.1
# Run the Dokploy deployment or restart procedure documented for your installation.
For a containerized deployment, update the image reference to the vendor-published image corresponding to 0.31.1 or later, then recreate the service using the deployment method used by your organization. Confirm that the new container is serving traffic and that old vulnerable containers are stopped.
If immediate patching is impossible, restrict access to Dokploy’s management interface through a VPN, identity-aware proxy, firewall policy, or equivalent control. Remove unnecessary permission to update database passwords, disable unused accounts, rotate credentials for accounts that may have been exposed, and monitor all password-change operations. These measures reduce opportunity but do not remove the vulnerability.
Technical Notes
A practical remediation sequence is:
# 1. Record the current deployment and running version
docker ps --format 'table {{.Names}}\t{{.Image}}'
docker inspect <dokploy-container> --format '{{.Config.Image}}'
# 2. Pull the approved fixed image or update the deployment source
docker pull <approved-dokploy-image>:0.31.1
# 3. Recreate the service using the organization’s deployment definition
docker compose up -d --force-recreate
# 4. Confirm the running image and service health
docker ps --format 'table {{.Names}}\t{{.Image}}\t{{.Status}}'
The placeholder image name is intentional: administrators must use the image repository and deployment method documented for their Dokploy installation. The essential requirement is that the running version is 0.31.1 or later, not merely that a local image was downloaded.
If exploitation is suspected, isolate the Dokploy host or container environment, preserve logs and process telemetry, rotate Dokploy, database, cloud, registry, SSH, and deployment credentials, and inspect managed workloads for tampering. Because the process commonly runs as root, assume that host-level secrets and connected systems may require review.
References
The primary vulnerability record is the NVD entry for CVE-2026-129987. It identifies Dokploy versions before 0.31.1 as affected and records the 9.9 Critical base score.
Technical and remediation references include the Dokploy patch commit, the related Dokploy pull request, and the Dokploy 0.31.1 release.
The NVD also lists the GitHub Security Advisory GHSA-9932-qg8x-pr32. For exploitation-status validation, consult the CISA Known Exploited Vulnerabilities catalog.
Because the supplied NVD result did not include the complete CVSS vector, defenders should verify the current NVD detail page before publishing the vector or incorporating individual metric values into risk-scoring workflows.
This article may contain affiliate links. We earn a commission on qualifying purchases at no extra cost to you.