Six minutes after uploading v0.1.0 of the malicious data-pipeline-check package, the attacker published v0.1.1 with byte-for-byte identical payload code. Armando's analysis confirms that both releases of this fake PyPI data pipeline validator contain the same import-time credential stealer, designed to silently scan developer home directories for cryptocurrency wallet keys, SSH credentials, and API tokens and exfiltrate them to an attacker-controlled endpoint. The package has been removed from PyPI; with 357 total downloads across both versions, any environment that imported it should be treated as compromised.
Key Findings
- Malicious version: v0.1.1, published 2026-05-24; no clean version exists -- v0.1.0, released six minutes earlier, carries the identical payload
- Entry point: module-level code in
_core.py fires a background thread automatically on import pipeline_check
- Targets: Python environments on any platform -- scans developer home directories for crypto and cloud credentials
- Evasion: 3-15 second random startup delay before scanning, to avoid triggering sandbox timing heuristics
- Detection: Text.Infostealer.SupplyChain (_core.py), Text.Trojan.SupplyChain (METADATA)
Package Background
data-pipeline-check presented itself as a utility to "validate data pipeline configurations and detect common misconfigurations." The description is plausible enough to attract engineers working with data engineering or MLOps tooling. The package was first published on May 24, 2026, and accumulated 357 total downloads before PyPI removed it. It declared no source code repository and no license -- red flags for any dependency review. The publisher account, DataOps on PyPI, published only this single package, and both versions it released were malicious.
Attack Analysis
The attack lives in pipeline_check/_core.py, a file whose docstring claims it is a "configuration validation engine." The module contains no actual validation logic. Instead, it builds a list of target directories under the victim's home directory: .ssh, .aws, .ethereum, .config, .docker, and .kube. The current working directory is also added to the scan list, capturing any credentials present in the project root at the time of import.
Before scanning, the attacker's code fetches the exfiltration endpoint from a GitHub Pages configuration file: https://ddjidd564.github.io/defi-security-best-practices/config.json. The URL mimics a legitimate security resource but delivers a JSON object containing attacker webhook URLs. SSL certificate verification is disabled in the request, ensuring delivery in any environment.
With webhook endpoints in hand, the stealer walks each target directory, reading files up to 2 MB each and applying six regex patterns to identify:
- 64-character hexadecimal strings matching cryptocurrency private keys
- 11-to-24 word sequences matching BIP39 wallet seed phrases
- OpenAI API keys (
sk-...)
- GitHub personal access tokens (
ghp_...)
- AWS access key IDs (
AKIA...)
- Plaintext passwords or passphrases in
KEY=value patterns
Matches are collected, capped at 200 findings, and POSTed as JSON to up to two webhook endpoints. Each payload includes hostname, platform, and current working directory alongside the credential fragments.
The stealer is invoked in a background daemon thread started at the moment pipeline_check is imported, via __init__.py. A random 3-to-15 second delay before the scan begins is a deliberate sandbox evasion measure -- the attacker chose this range to avoid detection systems that flag outbound connections made immediately on execution.
v0.1.1's _core.py is byte-for-byte identical to v0.1.0's. Publishing a second version six minutes after the first is consistent with an attacker ensuring the malicious release survives a potential early takedown of the first upload.
Malicious Files
_core.py is the complete attack payload, detected as Text.Infostealer.SupplyChain. It runs at import time via a background thread, scans six credential-rich directories and the current working directory, collects private keys, BIP39 mnemonics, and API tokens, fetches a remote webhook list from https://ddjidd564.github.io/defi-security-best-practices/config.json, and POSTs findings to those endpoints. The 3-15 second random startup delay is a sandbox evasion technique. The code is byte-for-byte identical to v0.1.0.
METADATA is the package distribution metadata file, detected as Text.Trojan.SupplyChain. It is a fingerprinted artifact of this malicious distribution that the RL scanner confirmed as a known malicious file.
Recommendations
Developers and security teams should take the following steps:
- Identify exposure: Check whether data-pipeline-check v0.1.1 appears in your dependency tree, including transitive dependencies. Any pipeline that ran
pip install while this version was current should be considered potentially affected.
- Update immediately: No clean version of this package exists. Remove data-pipeline-check from your dependencies entirely -- neither v0.1.0 nor v0.1.1 is safe.
- Inspect affected systems: On machines that imported this package, assume all credentials in
.ssh, .aws, .ethereum, .config, .docker, .kube, and any project working directory were scanned. Rotate crypto wallet keys, API tokens, GitHub tokens, AWS keys, and SSH keys immediately.
- Scan your software supply chain: Use Spectra Assure Community to check whether packages in your environment have been flagged.
Indicators of Compromise
Malicious Files
| Filename | SHA256 | Detection |
|---|
| METADATA | 227d318d0b542612e4e21b0971ccdd9e77a9d1af1b70e965a69ab1b7d9424ac6 | Text.Trojan.SupplyChain |
Network IOCs
| URL / IP | Purpose |
|---|
| https://ddjidd564.github.io/defi-security-best-practices/config.json | C2 configuration endpoint; returns JSON object with attacker webhook URLs used for credential exfiltration |
Archive Hashes
| File | SHA256 |
|---|
| data_pipeline_check-0.1.1-py3-none-any.whl | dd39f9db2b22fb97bbd26f64e46ce8df92d1899628e5c5f342a84774f3e8c76e |