qzc690385PyPI
Advisory published Updated
qzc690385 is a confirmed malicious PyPI package (MAL-2026-18751) that steals credentials and exfiltrates sensitive data (malicious versions 1.0.0, 1.0.1, 1.0.2…). Do not install it — remove it immediately and rotate any exposed credentials.
Malicious code in qzc690385 (PyPI)
What this malware does
At install time, setup.py polls the local Xinference API at http://127.0.0.1:9997/v1/model_registrations/flexible, selects model registrations whose names begin with 'xcmd0', url-safe-base64-decodes the suffix into a shell command, and runs it via subprocess.run(['/bin/sh', '-c',...]). The stdout/stderr of each execution is chunked, base64-encoded, and written back as new model registrations named 'zqx9p00<out>', 'zqx9p01<out>', 'zqx9err', with a 'zqx9alive' registration used as a liveness beacon. The command bytes are obfuscated as the url-safe-base64 suffix of a model name (reassembled by replacing '-'/'_' and padding '=' before base64 decode), so the payload only materializes after decode. The package has no declared purpose, an empty PKG-INFO, and a random-looking name, and ships no legitimate functionality. The effect is a bidirectional command-execution and output-exfiltration channel driven at pip install time: any party able to post model registrations to the installer's Xinference service (locally, or via SSRF/misconfiguration of that service) can run arbitrary shell commands on the installer's host as the installing user and read the results back through the same registry.
In this campaign, packages execute malicious code during installation. The exact behavior varies between versions: fingerprinting environment, exposing a backdoor server listening for remote commands, downloading remote commands from a hardcoded location. Next stages seem to be downloaded from localhost, what suggests the code may still be a preparation for the real attack.
Category: MALICIOUS - The campaign has clearly malicious intent, like infostealers.
Campaign: 2026-11-qzc
Reasons (based on the campaign):
-
The package contains code to exfiltrate basic data from the system, like IP or username. It has a limited risk.
-
The package contains code to execute remote commands (probably limited to a specific set) on the victim's machine.
-
backdoor
-
The package overrides the install command in setup.py to execute malicious code during installation.
Malicious versions
Indicators of compromise (SHA-256)
Detection & response playbook
Credential / info stealerFind it
Search your lockfiles and build artifacts for qzc690385 (14 malicious versions).
If you installed it — respond
qzc690385 is built to steal secrets, so assume every credential the build or runtime could read is compromised. Remove it from your project and lockfile, then rotate ALL exposed secrets — npm/registry tokens, cloud keys, CI/CD secrets, SSH keys, and any .env values — from a known-clean machine. Audit logs for unauthorized use of those credentials.
Did it already run?
If qzc690385 was installed, its post-install payload may already have run. Removing the package does not undo that — check outbound connections and credential use from the install window onward.
Frequently asked questions
Campaign
References
Credits
- Amazon Inspector · finder
- Kamil Mańkowski (kam193) · reporter
Detect & block this
O3 blocks qzc690385-class packages before install and in CI — and if it already ran, its runtime egress monitoring catches the credential exfiltration and severs the channel.