MAL-2026-13473 Malicious code in alphalend-layouts (PyPI)

The package harvests installer-side secrets and uploads them to an attacker-controlled GitHub repository, executing both when pip install alphalend-layouts runs (via custom install/develop cmdclass subclasses in setup.py that invoke _compat._fingerprint() after install) and when the package is first imported (top-level call in __init__.py). _candidate_files() reads ~/.sui/sui.keystore, ~/.sui/sui_config/*.yaml/client.yaml, and .env* files filtered by the markers SUI_, PK_B64, and PRIVATE_KEY. _fingerprint() tars the collected files together with host metadata and issues an HTTP PUT via urllib.request to https://api.github.com/repos/futongwan/sui-research-notes/contents/fingerprints/<ts>-<pid>.tar.gz. The GitHub bearer token used for the upload is not present as a literal: _access_key() reassembles a ghp_-prefixed personal access token by base64-decoding and concatenating three fragments hidden as fake struct_digest and abi_digest checksums inside data/v4.json plus a _COMPAT_TAG constant in _compat.py. Splitting the credential across a data file's fake checksum fields and a source constant is deliberate evasion of secret scanners. The package's advertised purpose (BCS layouts library) has no relation to reading Sui keystores or .env secrets or uploading archives to a third-party GitHub repository.

During import, the package exfiltrates sensitive files with SUI private keys to a private GitHub repository. This action is also triggered on every Python startup due to leveraging PTH files.

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

Campaign: 2026-08-alphalend-layouts

Reasons (based on the campaign):

• files-exfiltration

• obfuscation

• crypto-related

• exfiltration-crypto

• abuses-pth