Malicious code in dlmm-sdk (PyPI)

Overview

---

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

Source: amazon-inspector (9ddf2d03d839f7eff709e78da1d6a85226cdd3b9171dbcb59a4a95c741e0773e)

The package's top-level module runs a report() routine on import, and setup.py invokes the same import at install time. The routine collects environment variables matching secret-shaped keywords (KEY, TOKEN, SECRET, AWS, GITHUB, NPM, MNEMONIC, WALLET, SOLANA, HELIUS,...), reads ~/.npmrc and ~/.gitconfig, enumerates ~/.ssh, ~/.aws, ~/.config/solana, ~/.config/anchor, and ~/.config/gcloud, and gathers hostname, username, cwd, and platform. The collected JSON is POSTed via urllib.request.urlopen to a hardcoded webhook.site endpoint (https://webhook.site/326b0891-2093-4800-a4c1-686ce3e07b09). Module docstring and setup.py comment label the behavior 'environment diagnostics,' but the code path is bulk credential and host-identity exfiltration to a non-first-party collector. The package name resembles Meteora's DLMM SDK.

Source: kam193 (007be0fc2d53a2c72f277ddb12c24bb04ae1e17d2bf03f83b70367c3bf1b9122)

During import the package exfiltrates sensitive env variables and credential files. In addition, listings of cryptocurrency wallet directories are collected.

---

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

Campaign: 2026-08-dlmm

Reasons (based on the campaign):

• exfiltration-env-variables

• dependency-confusion

• exfiltration-credentials

• crypto-related

Affected packages

• dlmm-sdk

Remediation

Refer to the advisory for the patched release.