---
id: "MAL-2026-19952"
page: "https://infrastack.ai/live/MAL-2026-19952"
kind: "package"
aliases: ["MAL-2026-19952"]
title: "Malicious code in usdm-minting-client (PyPI)"
severity: "critical"
cvss: null
epss: null
exploited: false
status: "open"
category: "package"
products: [{"name": "usdm-minting-client", "fixed": [], "affected": "= 0.1.2; = 0.1.0; = 0.1.3; = 0.1.1", "ecosystem": "pypi"}]
ecosystems: ["pypi"]
tags: ["malicious", "package"]
not_affected_if: []
published_at: "2026-11-16T09:21:14+00:00"
first_seen_at: "2026-11-16T11:04:42+00:00"
updated_at: "2026-11-16T23:37:17+00:00"
sources: [{"source": "osv-malicious", "url": "https://osv.dev/vulnerability/MAL-2026-19952", "first_seen_at": "2026-11-16T11:04:42+00:00"}]
exploit_sources: []
---

# Malicious code in usdm-minting-client (PyPI)

Link for people: https://infrastack.ai/live/MAL-2026-19952 (this .md is for agents)

## Affected

- pypi:usdm-minting-client affected = 0.1.2; = 0.1.0; = 0.1.3; = 0.1.1 fixed no fixed version listed

## Timeline

- 2026-11-16T11:04:42+00:00 osv-malicious https://osv.dev/vulnerability/MAL-2026-19952

## From the sources

---
_-= Per source details. Do not edit below this line.=-_

## Source: amazon-inspector (3a76785d8f5b95d2ce973cec36ca5d1e7416205636b7d652a2fcd7f5124d7c01)
usdm-minting-client 0.1.1 performs credential theft at both install time and import time. setup.py defines a PostInstall cmdclass and also invokes an _exfil() function at module import of setup.py, so building the sdist via a default pip install causes the harvest to run. The function reads a curated set of wallet and cloud environment variables (PRIVATE_KEY, MNEMONIC, AWS_*, GOOGLE_APPLICATION_CREDENTIALS, UTILA_*, and similar) and walks the current working directory, INIT_CWD, $HOME, ~/.utila, and ~/.config for.env,.pem,.key,.json, and.yaml files, then POSTs the collected values as JSON to the URL configured in COLLECTOR_URL (default https://collector.example/collect) via urllib.request.urlopen. The package's top-level __init__.py repeats the exfiltration on import without any guard other than a ~/.usdm_client_exfil_done marker file: it reads USDM_PRIVATE_KEY, PRIVATE_KEY, MNEMONIC, and USDM_RPC_URL from the environment along with cwd/.env, ~/.env, and ~/.usdm.env, and POSTs them to the same collector. The package header self-labels as a 'PoC: usdm-minting-client (dependency confusion)' targeting Utila/USDm minting workflows, and the PKG-INFO lists a generic example.com homepage with a free-mail author. The self-label does not change the behavior: the shipped code is a working installer-side credential harvester that fires on every default pip install on the listed sdist-building platforms and on every import in production or CI.

## Source: kam193 (78ca43ab87b214f4da683e6b9f171e762e4956e6db3210796b6ed8bc92e92e72)
During installation the package exfiltrates credential files and environment variables, focusing on cryptocurrency- and cloud-related entries. The package seems to be targeting users of a package from GitHub (https://github.com/ethena-labs/usdm-minting-client) which has not been published to PyPI.


---

Category: MALICIOUS - The campaign has clearly malicious intent, like infostealers.


Campaign: 2026-11-usdm-minting-client


Reasons (based on the campaign):


 - The package overrides the install command in setup.py to execute malicious code during installation.


 - exfiltration-env-variables


 - crypto-related


 - exfiltration-crypto


 - exfiltration-credentials

---

_Free to read, with fair use of about 60 pages a minute. For higher limits or API access: sales@infrastack.ai_
